79
Distributed Detection and Estimation in Wireless Sensor Networks: Resource Allocation, Fusion Rules, and Network Security Edmond Nurellari The University of Leeds, UK School of Electronic and Electrical Engineering In accordance with the requirements for the degree of Doctor of Philosophy June 6, 2017 Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 1 / 42

Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

  • Upload
    others

  • View
    14

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Distributed Detection and Estimation in Wireless Sensor Networks:Resource Allocation, Fusion Rules, and Network Security

Edmond Nurellari

The University of Leeds, UKSchool of Electronic and Electrical Engineering

In accordance with the requirements for the degree of Doctor of Philosophy

June 6, 2017

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 1 / 42

Page 2: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Overview

1 Introduction

2 Optimal Quantization and Power Allocation

3 Centralized Quantized Fusion Rules

4 Distributed Two-Step Quantized Fusion Rules

5 Sensor Detection in the Presence of Falsified Observations

6 Summary

7 Key Conclusions

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 2 / 42

Page 3: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Overview

1 Introduction

2 Optimal Quantization and Power Allocation

3 Centralized Quantized Fusion Rules

4 Distributed Two-Step Quantized Fusion Rules

5 Sensor Detection in the Presence of Falsified Observations

6 Summary

7 Key Conclusions

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 2 / 42

Page 4: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Overview

1 Introduction

2 Optimal Quantization and Power Allocation

3 Centralized Quantized Fusion Rules

4 Distributed Two-Step Quantized Fusion Rules

5 Sensor Detection in the Presence of Falsified Observations

6 Summary

7 Key Conclusions

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 2 / 42

Page 5: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Overview

1 Introduction

2 Optimal Quantization and Power Allocation

3 Centralized Quantized Fusion Rules

4 Distributed Two-Step Quantized Fusion Rules

5 Sensor Detection in the Presence of Falsified Observations

6 Summary

7 Key Conclusions

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 2 / 42

Page 6: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Overview

1 Introduction

2 Optimal Quantization and Power Allocation

3 Centralized Quantized Fusion Rules

4 Distributed Two-Step Quantized Fusion Rules

5 Sensor Detection in the Presence of Falsified Observations

6 Summary

7 Key Conclusions

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 2 / 42

Page 7: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Overview

1 Introduction

2 Optimal Quantization and Power Allocation

3 Centralized Quantized Fusion Rules

4 Distributed Two-Step Quantized Fusion Rules

5 Sensor Detection in the Presence of Falsified Observations

6 Summary

7 Key Conclusions

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 2 / 42

Page 8: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Overview

1 Introduction

2 Optimal Quantization and Power Allocation

3 Centralized Quantized Fusion Rules

4 Distributed Two-Step Quantized Fusion Rules

5 Sensor Detection in the Presence of Falsified Observations

6 Summary

7 Key Conclusions

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 2 / 42

Page 9: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Overview

1 Introduction

2 Optimal Quantization and Power Allocation

3 Centralized Quantized Fusion Rules

4 Distributed Two-Step Quantized Fusion Rules

5 Sensor Detection in the Presence of Falsified Observations

6 Summary

7 Key Conclusions

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 3 / 42

Page 10: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

1. Introduction

Motivation

WSNs spatially deployed over a field can be designed to collect information and monitormany phenomena of interest.

Important role in several daily application scenarios such as health-care monitoring, homeapplications, smart farming, environment monitoring, and military.

1.2. Design Challenges in WSNs

1

FUSIONCENTERAttacker

. . .

SN1

SN2

SN3

SN4

SN5

SN6

SN7

SN8

SN9

SN10

Nature

T1

T2

T3

T4

T6

T5

SN1

SN3

SN10

SN8

←→−→

active sensor node manipulated by the attacker

active sensor node, working accordingly

non-active sensor node, manipulated by the attacker

non-active sensor node, controlled by the fusion center

partial inter-sensor node connectivity

target to SN link

SN1

SN3

SN10

SN8

←→−→

active sensor node, manipulated by the attacker

active sensor node, working accordingly

non-active sensor node, manipulated by the attacker

non-active sensor node, controlled by the fusion center

partial inter-sensor node connectivity

target to SN link

Figure 1.1: Schematic for a distributed communication architecture among periph-

eral SNs. Each SN generates a test statistic (Ti) by observing the target (thick lines).

The SNs have partial connectivity (thin lines) among themselves (i.e., not a complete

graph), but only over an energy-constrained/bandwidth-constrained network.

3

Figure 1: (left) A WSN architecture. (right) Smart city infrastructure.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 4 / 42

Page 11: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

1. Introduction

Motivation

WSNs spatially deployed over a field can be designed to collect information and monitormany phenomena of interest.

Important role in several daily application scenarios such as health-care monitoring, homeapplications, smart farming, environment monitoring, and military.

1.2. Design Challenges in WSNs

1

FUSIONCENTERAttacker

. . .

SN1

SN2

SN3

SN4

SN5

SN6

SN7

SN8

SN9

SN10

Nature

T1

T2

T3

T4

T6

T5

SN1

SN3

SN10

SN8

←→−→

active sensor node manipulated by the attacker

active sensor node, working accordingly

non-active sensor node, manipulated by the attacker

non-active sensor node, controlled by the fusion center

partial inter-sensor node connectivity

target to SN link

SN1

SN3

SN10

SN8

←→−→

active sensor node, manipulated by the attacker

active sensor node, working accordingly

non-active sensor node, manipulated by the attacker

non-active sensor node, controlled by the fusion center

partial inter-sensor node connectivity

target to SN link

Figure 1.1: Schematic for a distributed communication architecture among periph-

eral SNs. Each SN generates a test statistic (Ti) by observing the target (thick lines).

The SNs have partial connectivity (thin lines) among themselves (i.e., not a complete

graph), but only over an energy-constrained/bandwidth-constrained network.

3

Figure 1: (left) A WSN architecture. (right) Smart city infrastructure.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 4 / 42

Page 12: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

1. Introduction

Motivation

WSNs spatially deployed over a field can be designed to collect information and monitormany phenomena of interest.

Important role in several daily application scenarios such as health-care monitoring, homeapplications, smart farming, environment monitoring, and military.

1.2. Design Challenges in WSNs

1

FUSIONCENTERAttacker

. . .

SN1

SN2

SN3

SN4

SN5

SN6

SN7

SN8

SN9

SN10

Nature

T1

T2

T3

T4

T6

T5

SN1

SN3

SN10

SN8

←→−→

active sensor node manipulated by the attacker

active sensor node, working accordingly

non-active sensor node, manipulated by the attacker

non-active sensor node, controlled by the fusion center

partial inter-sensor node connectivity

target to SN link

SN1

SN3

SN10

SN8

←→−→

active sensor node, manipulated by the attacker

active sensor node, working accordingly

non-active sensor node, manipulated by the attacker

non-active sensor node, controlled by the fusion center

partial inter-sensor node connectivity

target to SN link

Figure 1.1: Schematic for a distributed communication architecture among periph-

eral SNs. Each SN generates a test statistic (Ti) by observing the target (thick lines).

The SNs have partial connectivity (thin lines) among themselves (i.e., not a complete

graph), but only over an energy-constrained/bandwidth-constrained network.

3

Figure 1: (left) A WSN architecture. (right) Smart city infrastructure.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 4 / 42

Page 13: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

1. Introduction

Design Challenges in WSNs

Low Power Hardware: Clearly, the biggest design constraint in WSNs still remains thepower consumption. Even-though the SNs are being designed using low-power microcontrollers, their power dissipation is still orders of magnitude too high.

Resource Constraints: Battery operated devices with limited on-board energy, both thesystem lifetime and communication bandwidth (BW) are restricted. Both the signalprocessing and communication should be carefully designed to consume minimal energy inorder to extend the lifetime and improve the overall reliability of the WSN.

Network Security:Usually unattended (geographically dispersed) and this makes themvulnerable to attacks. The overall detection and estimation strongly depends on thereliability of these SNs.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 5 / 42

Page 14: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

1. Introduction

Design Challenges in WSNs

Low Power Hardware: Clearly, the biggest design constraint in WSNs still remains thepower consumption. Even-though the SNs are being designed using low-power microcontrollers, their power dissipation is still orders of magnitude too high.

Resource Constraints: Battery operated devices with limited on-board energy, both thesystem lifetime and communication bandwidth (BW) are restricted. Both the signalprocessing and communication should be carefully designed to consume minimal energy inorder to extend the lifetime and improve the overall reliability of the WSN.

Network Security:Usually unattended (geographically dispersed) and this makes themvulnerable to attacks. The overall detection and estimation strongly depends on thereliability of these SNs.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 5 / 42

Page 15: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

1. Introduction

Design Challenges in WSNs

Low Power Hardware: Clearly, the biggest design constraint in WSNs still remains thepower consumption. Even-though the SNs are being designed using low-power microcontrollers, their power dissipation is still orders of magnitude too high.

Resource Constraints: Battery operated devices with limited on-board energy, both thesystem lifetime and communication bandwidth (BW) are restricted. Both the signalprocessing and communication should be carefully designed to consume minimal energy inorder to extend the lifetime and improve the overall reliability of the WSN.

Network Security:Usually unattended (geographically dispersed) and this makes themvulnerable to attacks. The overall detection and estimation strongly depends on thereliability of these SNs.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 5 / 42

Page 16: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Contribution-Publications List

1 E. Nurellari, D. McLernon, and M. Ghogho “A Secure Optimum Distributed Detection Scheme in Under-AttackWireless Sensor Networks”, in IEEE Trans. on Signal and Information Processing over Networks (TSIPN), April2017.

2 E. Nurellari, D. McLernon, and M. Ghogho, “Distributed Binary Event Detection Under Data-Falsification andEnergy-Bandwidth Limitation”, in IEEE Sensors Journal, vol. 16, no. 16, pp. 6298-6309, Aug. 15, 2016.

3 E. Nurellari, D. McLernon, and M. Ghogho, “Distributed Two-Step Quantized Fusion Rules via ConsensusAlgorithm for Distributed Detection in Wireless Sensor Networks,” in IEEE Transactions on Signal and InformationProcessing over Networks (TSIPN), vol. 2, no. 3, pp. 321-335, Sept. 2016.

4 S. Aldalahmeh, M. Ghogho, D. McLernon, and E. Nurellari, “Optimal fusion rule for distributed detection inclustered wireless sensor networks”, EURASIP Journal on Advances in Signal Process., 2016:5, Jan. 2016.

5 E. Nurellari, D. McLernon, and M. Ghogho, “Distributed detection in practical wireless sensor networks via a twostep consensus algorithm,” in Proc. IET Int. conf. on Intelligent Signal Process. (ISP), London, United Kingdom,1-2 Dec. 2015.

6 E. Nurellari, D. McLernon, M. Ghogho and S. A. R. Zaidi, “Distributed Optimal Quantization and PowerAllocation for Sensor Detection Via Consensus,” Proc. IEEE VTC Spring, Glasgow, U.K., 11-14 May 2015.

7 E. Nurellari, S. Aldalahmeh, M. Ghogho, and D. McLernon, “Quantized Fusion Rules for Energy-BasedDistributed Detection in Wireless Sensor Networks,” Proc. IEEE SSPD, Edinburgh, Scotland, 8-9 Sep. 2014.

8 E. Nurellari, D. McLernon, M. Ghogho and S. Aldalahmeh, “Optimal quantization and power allocation forenergy-based distributed sensor detection,” Proc. IEEE EUSIPCO, Lisbon, Portugal, 1-5 Sept. 2014.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 6 / 42

Page 17: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Contribution-Publications List

1 E. Nurellari, D. McLernon, and M. Ghogho “A Secure Optimum Distributed Detection Scheme in Under-AttackWireless Sensor Networks”, in IEEE Trans. on Signal and Information Processing over Networks (TSIPN), April2017.

2 E. Nurellari, D. McLernon, and M. Ghogho, “Distributed Binary Event Detection Under Data-Falsification andEnergy-Bandwidth Limitation”, in IEEE Sensors Journal, vol. 16, no. 16, pp. 6298-6309, Aug. 15, 2016.

3 E. Nurellari, D. McLernon, and M. Ghogho, “Distributed Two-Step Quantized Fusion Rules via ConsensusAlgorithm for Distributed Detection in Wireless Sensor Networks,” in IEEE Transactions on Signal and InformationProcessing over Networks (TSIPN), vol. 2, no. 3, pp. 321-335, Sept. 2016.

4 S. Aldalahmeh, M. Ghogho, D. McLernon, and E. Nurellari, “Optimal fusion rule for distributed detection inclustered wireless sensor networks”, EURASIP Journal on Advances in Signal Process., 2016:5, Jan. 2016.

5 E. Nurellari, D. McLernon, and M. Ghogho, “Distributed detection in practical wireless sensor networks via a twostep consensus algorithm,” in Proc. IET Int. conf. on Intelligent Signal Process. (ISP), London, United Kingdom,1-2 Dec. 2015.

6 E. Nurellari, D. McLernon, M. Ghogho and S. A. R. Zaidi, “Distributed Optimal Quantization and PowerAllocation for Sensor Detection Via Consensus,” Proc. IEEE VTC Spring, Glasgow, U.K., 11-14 May 2015.

7 E. Nurellari, S. Aldalahmeh, M. Ghogho, and D. McLernon, “Quantized Fusion Rules for Energy-BasedDistributed Detection in Wireless Sensor Networks,” Proc. IEEE SSPD, Edinburgh, Scotland, 8-9 Sep. 2014.

8 E. Nurellari, D. McLernon, M. Ghogho and S. Aldalahmeh, “Optimal quantization and power allocation forenergy-based distributed sensor detection,” Proc. IEEE EUSIPCO, Lisbon, Portugal, 1-5 Sept. 2014.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 6 / 42

Page 18: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Contribution-Publications List

1 E. Nurellari, D. McLernon, and M. Ghogho “A Secure Optimum Distributed Detection Scheme in Under-AttackWireless Sensor Networks”, in IEEE Trans. on Signal and Information Processing over Networks (TSIPN), April2017.

2 E. Nurellari, D. McLernon, and M. Ghogho, “Distributed Binary Event Detection Under Data-Falsification andEnergy-Bandwidth Limitation”, in IEEE Sensors Journal, vol. 16, no. 16, pp. 6298-6309, Aug. 15, 2016.

3 E. Nurellari, D. McLernon, and M. Ghogho, “Distributed Two-Step Quantized Fusion Rules via ConsensusAlgorithm for Distributed Detection in Wireless Sensor Networks,” in IEEE Transactions on Signal and InformationProcessing over Networks (TSIPN), vol. 2, no. 3, pp. 321-335, Sept. 2016.

4 S. Aldalahmeh, M. Ghogho, D. McLernon, and E. Nurellari, “Optimal fusion rule for distributed detection inclustered wireless sensor networks”, EURASIP Journal on Advances in Signal Process., 2016:5, Jan. 2016.

5 E. Nurellari, D. McLernon, and M. Ghogho, “Distributed detection in practical wireless sensor networks via a twostep consensus algorithm,” in Proc. IET Int. conf. on Intelligent Signal Process. (ISP), London, United Kingdom,1-2 Dec. 2015.

6 E. Nurellari, D. McLernon, M. Ghogho and S. A. R. Zaidi, “Distributed Optimal Quantization and PowerAllocation for Sensor Detection Via Consensus,” Proc. IEEE VTC Spring, Glasgow, U.K., 11-14 May 2015.

7 E. Nurellari, S. Aldalahmeh, M. Ghogho, and D. McLernon, “Quantized Fusion Rules for Energy-BasedDistributed Detection in Wireless Sensor Networks,” Proc. IEEE SSPD, Edinburgh, Scotland, 8-9 Sep. 2014.

8 E. Nurellari, D. McLernon, M. Ghogho and S. Aldalahmeh, “Optimal quantization and power allocation forenergy-based distributed sensor detection,” Proc. IEEE EUSIPCO, Lisbon, Portugal, 1-5 Sept. 2014.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 6 / 42

Page 19: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Contribution-Publications List

1 E. Nurellari, D. McLernon, and M. Ghogho “A Secure Optimum Distributed Detection Scheme in Under-AttackWireless Sensor Networks”, in IEEE Trans. on Signal and Information Processing over Networks (TSIPN), April2017.

2 E. Nurellari, D. McLernon, and M. Ghogho, “Distributed Binary Event Detection Under Data-Falsification andEnergy-Bandwidth Limitation”, in IEEE Sensors Journal, vol. 16, no. 16, pp. 6298-6309, Aug. 15, 2016.

3 E. Nurellari, D. McLernon, and M. Ghogho, “Distributed Two-Step Quantized Fusion Rules via ConsensusAlgorithm for Distributed Detection in Wireless Sensor Networks,” in IEEE Transactions on Signal and InformationProcessing over Networks (TSIPN), vol. 2, no. 3, pp. 321-335, Sept. 2016.

4 S. Aldalahmeh, M. Ghogho, D. McLernon, and E. Nurellari, “Optimal fusion rule for distributed detection inclustered wireless sensor networks”, EURASIP Journal on Advances in Signal Process., 2016:5, Jan. 2016.

5 E. Nurellari, D. McLernon, and M. Ghogho, “Distributed detection in practical wireless sensor networks via a twostep consensus algorithm,” in Proc. IET Int. conf. on Intelligent Signal Process. (ISP), London, United Kingdom,1-2 Dec. 2015.

6 E. Nurellari, D. McLernon, M. Ghogho and S. A. R. Zaidi, “Distributed Optimal Quantization and PowerAllocation for Sensor Detection Via Consensus,” Proc. IEEE VTC Spring, Glasgow, U.K., 11-14 May 2015.

7 E. Nurellari, S. Aldalahmeh, M. Ghogho, and D. McLernon, “Quantized Fusion Rules for Energy-BasedDistributed Detection in Wireless Sensor Networks,” Proc. IEEE SSPD, Edinburgh, Scotland, 8-9 Sep. 2014.

8 E. Nurellari, D. McLernon, M. Ghogho and S. Aldalahmeh, “Optimal quantization and power allocation forenergy-based distributed sensor detection,” Proc. IEEE EUSIPCO, Lisbon, Portugal, 1-5 Sept. 2014.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 6 / 42

Page 20: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Contribution-Publications List

1 E. Nurellari, D. McLernon, and M. Ghogho “A Secure Optimum Distributed Detection Scheme in Under-AttackWireless Sensor Networks”, in IEEE Trans. on Signal and Information Processing over Networks (TSIPN), April2017.

2 E. Nurellari, D. McLernon, and M. Ghogho, “Distributed Binary Event Detection Under Data-Falsification andEnergy-Bandwidth Limitation”, in IEEE Sensors Journal, vol. 16, no. 16, pp. 6298-6309, Aug. 15, 2016.

3 E. Nurellari, D. McLernon, and M. Ghogho, “Distributed Two-Step Quantized Fusion Rules via ConsensusAlgorithm for Distributed Detection in Wireless Sensor Networks,” in IEEE Transactions on Signal and InformationProcessing over Networks (TSIPN), vol. 2, no. 3, pp. 321-335, Sept. 2016.

4 S. Aldalahmeh, M. Ghogho, D. McLernon, and E. Nurellari, “Optimal fusion rule for distributed detection inclustered wireless sensor networks”, EURASIP Journal on Advances in Signal Process., 2016:5, Jan. 2016.

5 E. Nurellari, D. McLernon, and M. Ghogho, “Distributed detection in practical wireless sensor networks via a twostep consensus algorithm,” in Proc. IET Int. conf. on Intelligent Signal Process. (ISP), London, United Kingdom,1-2 Dec. 2015.

6 E. Nurellari, D. McLernon, M. Ghogho and S. A. R. Zaidi, “Distributed Optimal Quantization and PowerAllocation for Sensor Detection Via Consensus,” Proc. IEEE VTC Spring, Glasgow, U.K., 11-14 May 2015.

7 E. Nurellari, S. Aldalahmeh, M. Ghogho, and D. McLernon, “Quantized Fusion Rules for Energy-BasedDistributed Detection in Wireless Sensor Networks,” Proc. IEEE SSPD, Edinburgh, Scotland, 8-9 Sep. 2014.

8 E. Nurellari, D. McLernon, M. Ghogho and S. Aldalahmeh, “Optimal quantization and power allocation forenergy-based distributed sensor detection,” Proc. IEEE EUSIPCO, Lisbon, Portugal, 1-5 Sept. 2014.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 6 / 42

Page 21: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Contribution-Publications List

1 E. Nurellari, D. McLernon, and M. Ghogho “A Secure Optimum Distributed Detection Scheme in Under-AttackWireless Sensor Networks”, in IEEE Trans. on Signal and Information Processing over Networks (TSIPN), April2017.

2 E. Nurellari, D. McLernon, and M. Ghogho, “Distributed Binary Event Detection Under Data-Falsification andEnergy-Bandwidth Limitation”, in IEEE Sensors Journal, vol. 16, no. 16, pp. 6298-6309, Aug. 15, 2016.

3 E. Nurellari, D. McLernon, and M. Ghogho, “Distributed Two-Step Quantized Fusion Rules via ConsensusAlgorithm for Distributed Detection in Wireless Sensor Networks,” in IEEE Transactions on Signal and InformationProcessing over Networks (TSIPN), vol. 2, no. 3, pp. 321-335, Sept. 2016.

4 S. Aldalahmeh, M. Ghogho, D. McLernon, and E. Nurellari, “Optimal fusion rule for distributed detection inclustered wireless sensor networks”, EURASIP Journal on Advances in Signal Process., 2016:5, Jan. 2016.

5 E. Nurellari, D. McLernon, and M. Ghogho, “Distributed detection in practical wireless sensor networks via a twostep consensus algorithm,” in Proc. IET Int. conf. on Intelligent Signal Process. (ISP), London, United Kingdom,1-2 Dec. 2015.

6 E. Nurellari, D. McLernon, M. Ghogho and S. A. R. Zaidi, “Distributed Optimal Quantization and PowerAllocation for Sensor Detection Via Consensus,” Proc. IEEE VTC Spring, Glasgow, U.K., 11-14 May 2015.

7 E. Nurellari, S. Aldalahmeh, M. Ghogho, and D. McLernon, “Quantized Fusion Rules for Energy-BasedDistributed Detection in Wireless Sensor Networks,” Proc. IEEE SSPD, Edinburgh, Scotland, 8-9 Sep. 2014.

8 E. Nurellari, D. McLernon, M. Ghogho and S. Aldalahmeh, “Optimal quantization and power allocation forenergy-based distributed sensor detection,” Proc. IEEE EUSIPCO, Lisbon, Portugal, 1-5 Sept. 2014.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 6 / 42

Page 22: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Contribution-Publications List

1 E. Nurellari, D. McLernon, and M. Ghogho “A Secure Optimum Distributed Detection Scheme in Under-AttackWireless Sensor Networks”, in IEEE Trans. on Signal and Information Processing over Networks (TSIPN), April2017.

2 E. Nurellari, D. McLernon, and M. Ghogho, “Distributed Binary Event Detection Under Data-Falsification andEnergy-Bandwidth Limitation”, in IEEE Sensors Journal, vol. 16, no. 16, pp. 6298-6309, Aug. 15, 2016.

3 E. Nurellari, D. McLernon, and M. Ghogho, “Distributed Two-Step Quantized Fusion Rules via ConsensusAlgorithm for Distributed Detection in Wireless Sensor Networks,” in IEEE Transactions on Signal and InformationProcessing over Networks (TSIPN), vol. 2, no. 3, pp. 321-335, Sept. 2016.

4 S. Aldalahmeh, M. Ghogho, D. McLernon, and E. Nurellari, “Optimal fusion rule for distributed detection inclustered wireless sensor networks”, EURASIP Journal on Advances in Signal Process., 2016:5, Jan. 2016.

5 E. Nurellari, D. McLernon, and M. Ghogho, “Distributed detection in practical wireless sensor networks via a twostep consensus algorithm,” in Proc. IET Int. conf. on Intelligent Signal Process. (ISP), London, United Kingdom,1-2 Dec. 2015.

6 E. Nurellari, D. McLernon, M. Ghogho and S. A. R. Zaidi, “Distributed Optimal Quantization and PowerAllocation for Sensor Detection Via Consensus,” Proc. IEEE VTC Spring, Glasgow, U.K., 11-14 May 2015.

7 E. Nurellari, S. Aldalahmeh, M. Ghogho, and D. McLernon, “Quantized Fusion Rules for Energy-BasedDistributed Detection in Wireless Sensor Networks,” Proc. IEEE SSPD, Edinburgh, Scotland, 8-9 Sep. 2014.

8 E. Nurellari, D. McLernon, M. Ghogho and S. Aldalahmeh, “Optimal quantization and power allocation forenergy-based distributed sensor detection,” Proc. IEEE EUSIPCO, Lisbon, Portugal, 1-5 Sept. 2014.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 6 / 42

Page 23: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Contribution-Publications List

1 E. Nurellari, D. McLernon, and M. Ghogho “A Secure Optimum Distributed Detection Scheme in Under-AttackWireless Sensor Networks”, in IEEE Trans. on Signal and Information Processing over Networks (TSIPN), April2017.

2 E. Nurellari, D. McLernon, and M. Ghogho, “Distributed Binary Event Detection Under Data-Falsification andEnergy-Bandwidth Limitation”, in IEEE Sensors Journal, vol. 16, no. 16, pp. 6298-6309, Aug. 15, 2016.

3 E. Nurellari, D. McLernon, and M. Ghogho, “Distributed Two-Step Quantized Fusion Rules via ConsensusAlgorithm for Distributed Detection in Wireless Sensor Networks,” in IEEE Transactions on Signal and InformationProcessing over Networks (TSIPN), vol. 2, no. 3, pp. 321-335, Sept. 2016.

4 S. Aldalahmeh, M. Ghogho, D. McLernon, and E. Nurellari, “Optimal fusion rule for distributed detection inclustered wireless sensor networks”, EURASIP Journal on Advances in Signal Process., 2016:5, Jan. 2016.

5 E. Nurellari, D. McLernon, and M. Ghogho, “Distributed detection in practical wireless sensor networks via a twostep consensus algorithm,” in Proc. IET Int. conf. on Intelligent Signal Process. (ISP), London, United Kingdom,1-2 Dec. 2015.

6 E. Nurellari, D. McLernon, M. Ghogho and S. A. R. Zaidi, “Distributed Optimal Quantization and PowerAllocation for Sensor Detection Via Consensus,” Proc. IEEE VTC Spring, Glasgow, U.K., 11-14 May 2015.

7 E. Nurellari, S. Aldalahmeh, M. Ghogho, and D. McLernon, “Quantized Fusion Rules for Energy-BasedDistributed Detection in Wireless Sensor Networks,” Proc. IEEE SSPD, Edinburgh, Scotland, 8-9 Sep. 2014.

8 E. Nurellari, D. McLernon, M. Ghogho and S. Aldalahmeh, “Optimal quantization and power allocation forenergy-based distributed sensor detection,” Proc. IEEE EUSIPCO, Lisbon, Portugal, 1-5 Sept. 2014.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 6 / 42

Page 24: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Overview

1 Introduction

2 Optimal Quantization and Power Allocation

3 Centralized Quantized Fusion Rules

4 Distributed Two-Step Quantized Fusion Rules

5 Sensor Detection in the Presence of Falsified Observations

6 Summary

7 Key Conclusions

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 7 / 42

Page 25: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

2. Optimal Quantization and Power Allocation

System Architecture

SN3

SN2

SN5

SN1

SN4

Target

Fusion Center

T qf =

M∑i=1

αi [Ti ]QT3

T2

T5

T4

T1

[T3]Q

[T2]Q

[T5]Q[T4]Q

[T1]Q

Figure 2: Communication architecture between peripheral SNs and the FC. Each SN generates a test statistic byobserving the target and can communicate with the FC only over an energy-constrained/bandwidth-constrained link.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 8 / 42

Page 26: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

2. Simulation Results 1/2

1 2 3 4 5 6 7 8 9 100

0.5

1

1.5

sensor i

h2 i

1 2 3 4 5 6 7 8 9 100

2

4

sensor i

p i

1 2 3 4 5 6 7 8 9 100

2

4

sensor i

L i

Equal weighting in (3.3.4)

Optimum weighting in (3.3.4)

Figure 3: Equal weight (αi = 1√M,∀i) and optimal weight combining (α = αopt) transmit power and

channel quantization bits allocation for Pfa = 0.1, Pt = 10, U = 0.1, and M = 10.Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 9 / 42

Page 27: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

2. Simulation Results 2/2

0 0.2 0.4 0.6 0.8 10

0.2

0.4

0.6

0.8

1

Probability of false alarm, Pfa

Prob

abilit

yof

dete

ctio

n,P d

Optimal weight, N=100 samplesOptimal weight, N=300 samplesEqual weight, N=100 samplesEqual weight, N=300 samples

Figure 4: Receiver operating characteristic with Pt = 10, U = 0.1 and M = 10 for two differentweighting schemes.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 10 / 42

Page 28: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Overview

1 Introduction

2 Optimal Quantization and Power Allocation

3 Centralized Quantized Fusion Rules

4 Distributed Two-Step Quantized Fusion Rules

5 Sensor Detection in the Presence of Falsified Observations

6 Summary

7 Key Conclusions

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 11 / 42

Page 29: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

3. Simulation Results 1/3

-14 -12 -10 -8 -6 -40

0.2

0.4

0.6

0.8

1

Prob

abilit

yof

dete

ctio

n.P d

a

(dB)

Opt LRT-based

LRT-based in (4.4.8)

Opt lin comb in (4.4.9)

Eq LRT-based

Linear combi in (4.3.9)

Eq lin combining

Figure 5: Probability of detection (Pd ) versus the signal to noise ratio (ξa) for M = 20, N = 10,Pt = 10, Pfa = 0.1 and B = 0.5.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 12 / 42

Page 30: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

3. Simulation Results 2/3

0 20 40 60 80 100 1200

0.2

0.4

0.6

0.8

1

number of samples, N

Prob

abilit

yof

dete

ctio

n,P d

Opt fusion rule, Pt=102

Opt linear combining, Pt=102

Opt fusion rule, Pt=10-1

Opt linear combining, Pt=10-1

Figure 6: Probability of detection (Pd ) versus the number of samples (N) for M = 10 sensors,Pfa = 0.1, ξa = −8.5 dB and B = 1.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 13 / 42

Page 31: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

3. Simulation Results 3/3

20 40 60 80 100

0.4

0.5

0.6

0.7

0.8

0.9

1

number of sensors, M

Prob

abilit

yof

dete

ctio

n,P d

Optimum fusion rule LRT-based

LRT-based with weights in (4.4.8)

Optimum linear combining in (4.4.9)

Equal weight LRT-based

Linear combining with weights in (4.3.9)

Equal weight linear combining

Figure 7: Probability of detection (Pd ) versus number of sensors (M) for N = 10, Pt = 10, Pfa = 0.1,ξa = −8.5 dB and B = 0.5.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 14 / 42

Page 32: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Overview

1 Introduction

2 Optimal Quantization and Power Allocation

3 Centralized Quantized Fusion Rules

4 Distributed Two-Step Quantized Fusion Rules

5 Sensor Detection in the Presence of Falsified Observations

6 Summary

7 Key Conclusions

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 15 / 42

Page 33: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

4. Distributed Two-Step Quantized Fusion Rules

Communication Architecture

SN1

SN2

SN3

SN5

SN4

SN6

Target

T1

T2

T3

T4

T6

T5

Figure 8: A distributed communication architecture among peripheral SNs. The SNs have partial connectivity (thinlines) among themselves (i.e., not a complete graph).

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 16 / 42

Page 34: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

4. Quantized Distributed Soft Decision Fusion Rule

Proposition

Here we propose a scheme, where SN i encodes the data (using a simple uniformquantizer with qi bits) prior to information exchange.

1 We also propose to establish a link between any two SNs i and j based on the (known) SNRat node j , i.e.

if SNRij < Υ, eij = eji = 0

if SNRij ≥ Υ, eij = eji = 1.

}

2 Υ is a SNR threshold parameter and SNRij defined as:

SNRij =pt

ijh2ij

ζ0dγij

.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 17 / 42

Page 35: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

4. Quantized Distributed Soft Decision Fusion Rule

Proposition

Here we propose a scheme, where SN i encodes the data (using a simple uniformquantizer with qi bits) prior to information exchange.

1 We also propose to establish a link between any two SNs i and j based on the (known) SNRat node j , i.e.

if SNRij < Υ, eij = eji = 0

if SNRij ≥ Υ, eij = eji = 1.

}

2 Υ is a SNR threshold parameter and SNRij defined as:

SNRij =pt

ijh2ij

ζ0dγij

.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 17 / 42

Page 36: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

4. Quantized Distributed Soft Decision Fusion Rule

Proposition

Here we propose a scheme, where SN i encodes the data (using a simple uniformquantizer with qi bits) prior to information exchange.

1 We also propose to establish a link between any two SNs i and j based on the (known) SNRat node j , i.e.

if SNRij < Υ, eij = eji = 0

if SNRij ≥ Υ, eij = eji = 1.

}

2 Υ is a SNR threshold parameter and SNRij defined as:

SNRij =pt

ijh2ij

ζ0dγij

.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 17 / 42

Page 37: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

4. Quantized Distributed Soft Decision Fusion Rule

Proposition

We propose to quantize with qi bits at SN i before transmitting to SN j :

qi ≤1

2log2 (1 + Υ) bits/sample

A large Υ means:

1 Fewer communication links and so slower information diffusion across the network.2 An increase in the number of bits that each SN can transmit to its neighbors.

A small Υ means:

1 Establishes a more connected graph and dictates a faster information diffusion across thenetwork.

2 Allows less transmission bits resulting in an increase in the quantization noise variance.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 18 / 42

Page 38: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

4. Quantized Distributed Soft Decision Fusion Rule

Proposition

We propose to quantize with qi bits at SN i before transmitting to SN j :

qi ≤1

2log2 (1 + Υ) bits/sample

A large Υ means:1 Fewer communication links and so slower information diffusion across the network.

2 An increase in the number of bits that each SN can transmit to its neighbors.

A small Υ means:

1 Establishes a more connected graph and dictates a faster information diffusion across thenetwork.

2 Allows less transmission bits resulting in an increase in the quantization noise variance.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 18 / 42

Page 39: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

4. Quantized Distributed Soft Decision Fusion Rule

Proposition

We propose to quantize with qi bits at SN i before transmitting to SN j :

qi ≤1

2log2 (1 + Υ) bits/sample

A large Υ means:1 Fewer communication links and so slower information diffusion across the network.2 An increase in the number of bits that each SN can transmit to its neighbors.

A small Υ means:

1 Establishes a more connected graph and dictates a faster information diffusion across thenetwork.

2 Allows less transmission bits resulting in an increase in the quantization noise variance.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 18 / 42

Page 40: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

4. Quantized Distributed Soft Decision Fusion Rule

Proposition

We propose to quantize with qi bits at SN i before transmitting to SN j :

qi ≤1

2log2 (1 + Υ) bits/sample

A large Υ means:1 Fewer communication links and so slower information diffusion across the network.2 An increase in the number of bits that each SN can transmit to its neighbors.

A small Υ means:1 Establishes a more connected graph and dictates a faster information diffusion across the

network.

2 Allows less transmission bits resulting in an increase in the quantization noise variance.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 18 / 42

Page 41: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

4. Quantized Distributed Soft Decision Fusion Rule

Proposition

We propose to quantize with qi bits at SN i before transmitting to SN j :

qi ≤1

2log2 (1 + Υ) bits/sample

A large Υ means:1 Fewer communication links and so slower information diffusion across the network.2 An increase in the number of bits that each SN can transmit to its neighbors.

A small Υ means:1 Establishes a more connected graph and dictates a faster information diffusion across the

network.2 Allows less transmission bits resulting in an increase in the quantization noise variance.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 18 / 42

Page 42: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

4. Simulation Results 1/6

0 10 20 30 40 50 60 70 800

0.5

1

(

Norm

.E# P T

$Proposed two-step

Conventional cons.

0 10 20 30 40 50 60 70 800.68

0.7

0.72

0.74

(

P$ d

0 10 20 30 40 50 60 70 800

0.5

1

(

;

Figure 9: Normalized average power consumption (E[PT

]), achievable8 probability of detection (P∗

d ) and the average

communication link density (ρ) versus Υ, with σ2eh

= 0, decision fusion in (5.4.16), Pgfa = 0.1, U = 3, N = 20, M = 17

and with αi (scaled by M).Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 19 / 42

Page 43: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

4. Simulation Results 2/6

0 0.2 0.4 0.6 0.8 10

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Global prob. of false alarm, P gfa

Glo

balp

rob.

ofde

tect

ion,

Pg d

Centralized optimum linear rule (5.3.12)Proposed weighted two-step, K1 = 100Proposed weighted two-step, K1 = 200Proposed weighted two-step, K1 = 350Proposed weighted two-step, K1 = 500Proposed weighted two-step, K1 = 800Centralized LRT-based [36]

K1 = 350K1 = 200

K1 = 100

K1 = 800K1 = 500

Upper bound

Figure 10: Averaged (over 500 h2ij realizations) ROC for the proposed two-step weighted algorithm with decision fusion

in (40), U = 3, N = 20, M = 17, K2 = 3, Υ = 30, σ2eh

= 0 and with αi (scaled by M) in (5.3.9).

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 20 / 42

Page 44: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

4. Simulation Results 3/6

0 0.2 0.4 0.6 0.8 10

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Global prob. of false alarm, P gfa

Glo

balp

rob.

ofde

tect

ion,

Pg d

Proposed weighted two-step, K1 = 50Proposed weighted two-step, K1 = 150Proposed weighted two-step, K1 = 300Centralized LRT-based [36]Centralized optimum linear rule (5.3.12)

K1 = 300

K1 = 150

Upper bound

K1 = 50

Figure 11: Averaged (over 500 h2ij realizations) ROC against first step iterations number (K1), with decision fusion in

(41), K2 = 2, U = 3, N = 20, M = 17, Υ = 10, σ2eh

= 0 and with αi (scaled by M) in (5.3.9).

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 21 / 42

Page 45: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

4. Simulation Results 4/6

-25 -20 -15 -10 -5 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

a

(dB)

Glo

balpro

b.

of

dete

ction,

Pg d

Centr. LRT-based in [36]Centr. opt linear rule (5.3.12)Proposed two-step with (5.4.16)Proposed two-step with (5.4.15)

LRT-based

Centr opt linear rule

Proposed with (5.4.15)

Proposed with (5.4.16)

-20 -15 -10 -5 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

a

(dB)

Glo

balpro

b.

of

dete

ction,

Pg d

Centr. LRT-based in [36]Proposed two-step with (5.4.16)Proposed two-step with (5.4.16)Proposed two-step with (5.4.16)

<2eh

= 4

<2eh

= 1<2

eh= 0

Figure 12: Averaged (over 500 h2ij realizations) probability of detection (Pg

d ) against the signal to noise ratio(ξa) with Pg

fa = 0.1, U = 3, N = 20, M = 17, K1 = 320, Υ = 20, ξi = ξ,∀i in (4) and with αi (scaled by M) in(5.3.9): (left) ideal, σ2

eh= 0; (right) non-ideal, σ2

eh6= 0.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 22 / 42

Page 46: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

4. Simulation Results 5/6

0 0.2 0.4 0.6 0.8 10

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Global prob. of false alarm, P gfa

Glo

balp

rob.

ofde

tect

ion,

Pg d

Proposed weighted two-step with (5.4.15)

Unquantized eq. comb. (,i = 1) in (5.3.14)

Proposed eq. comb. (,i = 1) two-step with (5.4.16)

Proposed eq. comb. (,i = 1) two-step with (5.4.15)

Proposed weighted two-step with (5.4.16)

Centr. opt. linear rule (5.3.12)

K1 = 400

K1 = 320

Centr. opt.

K1 = 320

Centr. eq. comb.

Figure 13: Averaged (over 500 h2ij realizations) ROC for the proposed (quantized) two-step weighted fusion rule

with U = 3, N = 20, Υ = 20, M = 17 and with αi (scaled by M) in (5.3.9).Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 23 / 42

Page 47: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

4. Simulation Results 6/6

-14 -12 -10 -8 -6 -4 -20

0.2

0.4

0.6

0.8

1

9a (dB)

Glo

balp

rob

ofde

tect

ion,

Pg d

Unquantized eq comb in (5.3.14)Proposed eq comb two-step, K1 = 10Proposed eq omb two-step, K1 = 20SN 3 eq comb -rst step, K1 = 10

K1 = 20

K1 = 10

Centralized detector

K1 = 10

Figure 14: Probability of detection (Pgd ) versus the signal to noise ratio (ξa) for M = 13, Υ = 72, U = 2, N = 20,

Pgfa = 0.1 and ξi = ξ.∀i in (3.2.4) and αi = 1, ∀i in (5.4.4). The topology used is given in right of Fig. 5.5.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 24 / 42

Page 48: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Overview

1 Introduction

2 Optimal Quantization and Power Allocation

3 Centralized Quantized Fusion Rules

4 Distributed Two-Step Quantized Fusion Rules

5 Sensor Detection in the Presence of Falsified Observations

6 Summary

7 Key Conclusions

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 25 / 42

Page 49: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

5. Sensor Detection in the Presence of Falsified Observations

Motivation1 Geographically dispersed to cover large areas, the SNs are constrained in both bandwidth

and power. Usually unattended and this makes them vulnerable to different attacks.

2 The overall detection performance strongly depends on the reliability of these SNs in thenetwork.

3 While fusing the data received by the spatially deployed SNs allows the FC to make areliable decision, it is possible that one or more SNs (compromised by an attacker)deliberately falsify their local observations.

Contributions

1 The problem of centralized detection in the presence of compromised SNs is investigated.

2 Attacker-based and FC-based parameter optimization are considered and some expressionshave been derived.

3 A reputation based scheme to identify the compromised SNs in the network and controltheir influence to the global FC decision is also proposed.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 26 / 42

Page 50: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

5. Sensor Detection in the Presence of Falsified Observations

Motivation1 Geographically dispersed to cover large areas, the SNs are constrained in both bandwidth

and power. Usually unattended and this makes them vulnerable to different attacks.

2 The overall detection performance strongly depends on the reliability of these SNs in thenetwork.

3 While fusing the data received by the spatially deployed SNs allows the FC to make areliable decision, it is possible that one or more SNs (compromised by an attacker)deliberately falsify their local observations.

Contributions

1 The problem of centralized detection in the presence of compromised SNs is investigated.

2 Attacker-based and FC-based parameter optimization are considered and some expressionshave been derived.

3 A reputation based scheme to identify the compromised SNs in the network and controltheir influence to the global FC decision is also proposed.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 26 / 42

Page 51: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

5. Sensor Detection in the Presence of Falsified Observations

Motivation1 Geographically dispersed to cover large areas, the SNs are constrained in both bandwidth

and power. Usually unattended and this makes them vulnerable to different attacks.

2 The overall detection performance strongly depends on the reliability of these SNs in thenetwork.

3 While fusing the data received by the spatially deployed SNs allows the FC to make areliable decision, it is possible that one or more SNs (compromised by an attacker)deliberately falsify their local observations.

Contributions

1 The problem of centralized detection in the presence of compromised SNs is investigated.

2 Attacker-based and FC-based parameter optimization are considered and some expressionshave been derived.

3 A reputation based scheme to identify the compromised SNs in the network and controltheir influence to the global FC decision is also proposed.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 26 / 42

Page 52: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

5. Sensor Detection in the Presence of Falsified Observations

Motivation1 Geographically dispersed to cover large areas, the SNs are constrained in both bandwidth

and power. Usually unattended and this makes them vulnerable to different attacks.

2 The overall detection performance strongly depends on the reliability of these SNs in thenetwork.

3 While fusing the data received by the spatially deployed SNs allows the FC to make areliable decision, it is possible that one or more SNs (compromised by an attacker)deliberately falsify their local observations.

Contributions1 The problem of centralized detection in the presence of compromised SNs is investigated.

2 Attacker-based and FC-based parameter optimization are considered and some expressionshave been derived.

3 A reputation based scheme to identify the compromised SNs in the network and controltheir influence to the global FC decision is also proposed.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 26 / 42

Page 53: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

5. Sensor Detection in the Presence of Falsified Observations

Motivation1 Geographically dispersed to cover large areas, the SNs are constrained in both bandwidth

and power. Usually unattended and this makes them vulnerable to different attacks.

2 The overall detection performance strongly depends on the reliability of these SNs in thenetwork.

3 While fusing the data received by the spatially deployed SNs allows the FC to make areliable decision, it is possible that one or more SNs (compromised by an attacker)deliberately falsify their local observations.

Contributions1 The problem of centralized detection in the presence of compromised SNs is investigated.

2 Attacker-based and FC-based parameter optimization are considered and some expressionshave been derived.

3 A reputation based scheme to identify the compromised SNs in the network and controltheir influence to the global FC decision is also proposed.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 26 / 42

Page 54: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

5. Sensor Detection in the Presence of Falsified Observations

Motivation1 Geographically dispersed to cover large areas, the SNs are constrained in both bandwidth

and power. Usually unattended and this makes them vulnerable to different attacks.

2 The overall detection performance strongly depends on the reliability of these SNs in thenetwork.

3 While fusing the data received by the spatially deployed SNs allows the FC to make areliable decision, it is possible that one or more SNs (compromised by an attacker)deliberately falsify their local observations.

Contributions1 The problem of centralized detection in the presence of compromised SNs is investigated.

2 Attacker-based and FC-based parameter optimization are considered and some expressionshave been derived.

3 A reputation based scheme to identify the compromised SNs in the network and controltheir influence to the global FC decision is also proposed.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 26 / 42

Page 55: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

5. Sensor Detection in the Presence of Falsified Observations

Communication Architecture

SN3

SN2

SN5

SN1

SN4

SN6

Target

Attacker

Fusion Center

Tf =M∑i=1

αiTqi

T fal3

T2

T fal5

T4

T6

T1

T q3

T q2

T q5

T q4

T q1

T q6

Figure 15: Under attack communication architecture between peripheral SNs and the FC. While the honest SNs teststatistics remain unchanged, the compromised SNs falsify their test statistics before transmitting to the FC.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 27 / 42

Page 56: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

5. Simulation Results 1/4

1 2 3 4 5 6 7 8 9 10 11 120

1

2

SNs

h i2

1 2 3 4 5 6 7 8 9 10 11 120

10

20

SNs

po i

1 2 3 4 5 6 7 8 9 10 11 120

5

SNs

L i

C=5C=0.5C=0

Figure 16: SN optimal transmit power (poi ) and channel bit allocation (Li ) with Pt = 60, U = 3,

ξa = −10.5 dB, N = 20, β = 0.1 and σ2eh

= 0.Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 28 / 42

Page 57: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

5. Simulation Results 2/4

0 0.2 0.4 0.6 0.8 10

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Probability of false alarm, Pfa

Prob

abili

tyof

dete

ction

,Pd

AF in [35]

opt. in (6.2.22), - = 0:1

OAFBB, - = 0:1

WAFBB, - = 0:1

opt. in (6.2.22), - = 0:5

OAFBB, - = 0:5

WAFBB, - = 0:5

OAFBB, - = 1

WAFBB, - = 1

=0.1

=0.5

=1

Figure 17: Probability of detection (Pd ) versus probability of false alarm (Pfa) with U = 3, Pt = 60,M = 12, N = 20, Ci = 0.9,∀i and σ2

eh= 0.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 29 / 42

Page 58: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

5. Simulation Results 3/4

0 0.2 0.4 0.6 0.8 10

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Probability of false alarm, Pfa

Prob

abili

tyof

dete

ctio

n,P d

C=0C=0.2C=0.4C=0.45C=0.6C=0.9C=1.1C=1.4

0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1

0.86

0.88

0.9

0.92

0.94

0.96

0.98

C=1.4 C=1.1

C=0.9

C=0.2

C=0.45

C=0

C=0.6

Nash Equilibrium, C=0.4

Figure 18: Probability of detection (Pd ) versus probability of false alarm (Pfa), with U = 3, ξa = −10.5dB, Pt = 60, M = 12, N = 20, β = 0.2, σ2

eh= 0 and with optimum weights in (6.2.22).

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 30 / 42

Page 59: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

5. Simulation Results 4/4

0 2 4 6 8 10-1

-0.5

0

0.5

1

1.5

2

2.5

3

Attacker strength, C

Modi-

edde.

ection

coe/c

ient,

~ d2Optimum , in (6.2.22)Non-optimum ,First derivative

Figure 19: Modified deflection coefficient (d2) versus the attacker strength (C) with U = 3, ξa = −10 dB,si = 0.1, ∀i , Pt = 60, M = 12, N = 20, β = 0.1 and σ2

eh= 0.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 31 / 42

Page 60: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

5. A Secure Sub-optimum Detection Scheme in Under-Attack WSNs

SN3

SN2

SN5

SN1

SN4

SN6

Target

Attacker

Fusion Center

Tf =M∑i=1

αi Ii

I C3

I2

I C5

I4

I6

I1

I3

I2

I5

I4

I1

I6

Figure 20: Under attack schematic communication architecture between peripheral SNs and the fusion center (FC).While the i th (i = {1, 2, 4, 6}) honest SN indicator (test statistic) remains unchanged (i.e., Ii = Ii ), the j th (j = {3, 5})compromised SN falsify its indicator (test statistic) as in (6.3.7) before transmitting to the FC.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 32 / 42

Page 61: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

5. A Secure Sub-optimum Detection Scheme in Under-Attack WSNs

FC Optimum Weighting

αiopt =

(1− β

)(pi

d − pifa

)+ β

(pi ,C

fa − pi ,Cd

)(2P fal

C − 1)(

1−β)(pi

d

(1−pi

d

))+β(Pflip

C + pi ,Cd

(1−2Pflip

C

))(1−Pflip

C + pi ,Cd

(2Pflip

C −1)) . (1)

Depends upon the local pifa and the pi

d as well as on the β (fraction of compromised SNs) andthe probability of flipping the local decisions by the attacker. The FC cannot implement theoptimum weight combining fusion ruleAttacker Flipping Probability Optimisation

Lemma 6.3.2: The optimum flipping probability(Pflip

C ,opt

)which minimizes the modified

deflection coefficient is:

PflipC ,opt =

β − 1

( M∑i=1

αi

(pi

d − pifa

)M∑

i=1αi

(pi ,C

fa − pi ,Cd

))

+1

2(2)

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 33 / 42

Page 62: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

5. Simulation Results 1/6

5 10 15 200

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

FC threshold, Λf

Repu

tation

metr

ic, r i

SN9SN10SN11SN12SN13SN15

5 10 15 200

0.05

0.1

0.15

0.2

0.25

0.3

0.35

FC threshold, Λf

Repu

tation

metr

ic, r i

SN21SN22SN23SN24SN29SN30SN32SN35SN39

Figure 21: The reliability metric (ri ) versus the FC detection threshold (Λf ) against the SNs with M = 40, N = 20,

β = 0.5, PflipC = 1 and K = 150.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 34 / 42

Page 63: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

5. Simulation Results 2/6

0 5 10 15 200

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

FC detection threshold, f

Prob

.ofd

et.t

heco

mpr

omise

dSN

37,P

37,t

rue

d

K=5

K=10

K=15

K=20

K=30

K=100

K=200

Figure 22: Probability that the (compromised) SN 37 has been truly detected (P37,trued ) versus the FC detection

threshold (Λf ) with M = 40, N = 20, β = 0.5, PflipC = 1 and δ = 0.009.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 35 / 42

Page 64: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

5. Simulation Results 3/6

5 10 15 20 25 30 35 40 45 500

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Time window length, K

Aver.

prob

.of

det.

(mis-det.)

,Ptrue

d(P

false

d)

P trued , Λf = 5

Pf alse

d , Λf = 5

P trued , Λf = 13

Pf alse

d , Λf = 13

P trued , Λf = 13

Pf alse

d , Λf = 13

β=0.5

β=0.25

β=0.10

Figure 23: Average compromised SNs detection probability and honest SNs mis-detection probability versus the time

window length (K) and against β with M = 40, N = 20, PflipC = 1 and δ = 0.009.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 36 / 42

Page 65: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

5. Simulation Results 4/6

1 2 3 4 5 6 7 8 9 100.2

0.25

0.3

0.35

0.4

0.45

0.5

0.55

0.6

0.65

Time window length, K

Prob

.ofd

et.m

inus

prob

.off

alse

alar

m,

P d!

P fa

Scheme in [73], $f = 7Scheme in [73], $f = 9Proposed, $f = 7Proposed, $f = 9

f=9

f=7

Figure 24: The Pd − Pfa metric versus the time window length (K) against the FC detection threshold (Λf ) with

M = 40, N = 20, β = 0.25, PflipC = 0.2, δ = 0.95 and µ = 10.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 37 / 42

Page 66: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

5. Simulation Results 5/6

0 0.2 0.4 0.6 0.8 10

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Prob

.of

dete

ctio

n,P d

Prob. of false alarm, Pfa

Upper Bound

Equal combining in [35]

Perfect SNs iden. and tot. removal

Opt. weights (6.3.24), perfect SNs iden.

Proposed, 7 = 15, / = 0:09

Proposed, 7 = 15, / = 0:12

Proposed, 7 = 55, / = 0:009

,i = ,AFi in (6.3.10), no iden. scheme

Proposed, 7 = 35, / = 0:009

Proposed

No ident. scheme

Figure 25: Probability of detection (Pd ) versus probability of false alarm (Pfa) with M = 40, N = 20, β = 0.5, PflipC = 1

and K = 5.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 38 / 42

Page 67: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

5. Simulation Results 6/6

0 0.2 0.4 0.6 0.8 10

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Prob. of false alarm, Pfa

Prob

.of

dete

ctio

n,P d Upper Bound

Opt. weights (6.3.24), perf. SNs iden.

Equal combining in [35], - = 0

Proposed, / = 0:009, K=80, 7 = 6

Proposed, / = 0:009, K=40, 7 = 6

Proposed, / = 0:009, K=5, 7 = 10

Proposed, / = 0:009, K=5, 7 = 14

Proposed, / = 0:009, K=5, 7 = 1

With ,i = ,AFi in (6.3.10), no iden.

Scheme in [73], / = 1, K = 5

Equal combining, - = 0:25

Proposed

Eq. comb.

No iden. scheme

Scheme in [73]

Figure 26: Probability of detection (Pd ) versus probability of false alarm (Pfa) against δ and µ with M = 40, N = 20,

β = 0.25, and PflipC = 1.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 39 / 42

Page 68: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Summary

We derive the optimum fusion rule and then analyze sub-optimum fusion rules that arerealizable and easily implemented in practical WSN deployment scenarios. The effect offading channels on detection performance is minimized by solving the resource allocationproblem.

A two-step consensus-based approach with weight combining quantized test statisticsexchange is proposed. We relate the communication topology with the number of bits tobe shared among SNs. It turns out that there is an optimum topology that maximizes thedetection performance.

Centralized detection in the presence of compromised SNs is also investigated. Attackerand FC based parameter optimization are considered and some expressions have beenderived. A reputation based scheme to identify the compromised SNs in the network andcontrol their influence to the global FC decision is also proposed.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 40 / 42

Page 69: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Summary

We derive the optimum fusion rule and then analyze sub-optimum fusion rules that arerealizable and easily implemented in practical WSN deployment scenarios. The effect offading channels on detection performance is minimized by solving the resource allocationproblem.

A two-step consensus-based approach with weight combining quantized test statisticsexchange is proposed. We relate the communication topology with the number of bits tobe shared among SNs. It turns out that there is an optimum topology that maximizes thedetection performance.

Centralized detection in the presence of compromised SNs is also investigated. Attackerand FC based parameter optimization are considered and some expressions have beenderived. A reputation based scheme to identify the compromised SNs in the network andcontrol their influence to the global FC decision is also proposed.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 40 / 42

Page 70: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Summary

We derive the optimum fusion rule and then analyze sub-optimum fusion rules that arerealizable and easily implemented in practical WSN deployment scenarios. The effect offading channels on detection performance is minimized by solving the resource allocationproblem.

A two-step consensus-based approach with weight combining quantized test statisticsexchange is proposed. We relate the communication topology with the number of bits tobe shared among SNs. It turns out that there is an optimum topology that maximizes thedetection performance.

Centralized detection in the presence of compromised SNs is also investigated. Attackerand FC based parameter optimization are considered and some expressions have beenderived. A reputation based scheme to identify the compromised SNs in the network andcontrol their influence to the global FC decision is also proposed.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 40 / 42

Page 71: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Key Conclusions

Shown that spatially distributed SNs across the field can offer a reliable operation for eventdetection applications. The system detection performance and the WSN’s operating lifetime canbe further improved by means of resource allocations, optimisation and signal processingalgorithms

=⇒ complexity to be kept as simple as possible.

The data fusion problem: we derive the optimal fusion rules (i.e., for attack-free and under-attackWSN scenarios) and have shown that these fusion rules are not implementable in practice andrequire complex local signal processing =⇒ Derive sub-optimum but simple fusion rules (requiringsimple hardware) that offer reliable and good detection performance.

A better but more complex approach is to possibly identify these compromised SNs and controltheir influence on the FC decision =⇒ Offers an improved detection performance but requiresobserving the SN’s local reports for a period of time. A larger observation time period (K) maylead to a large detection delay that is critical for most of the event detection applications.

We have addressed the fully distributed detection problem and proposed signal processingalgorithms for such an approach =⇒ Very attractive from both the signal processing perspectiveand the communication point of view.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 41 / 42

Page 72: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Key Conclusions

Shown that spatially distributed SNs across the field can offer a reliable operation for eventdetection applications. The system detection performance and the WSN’s operating lifetime canbe further improved by means of resource allocations, optimisation and signal processingalgorithms =⇒ complexity to be kept as simple as possible.

The data fusion problem: we derive the optimal fusion rules (i.e., for attack-free and under-attackWSN scenarios) and have shown that these fusion rules are not implementable in practice andrequire complex local signal processing =⇒ Derive sub-optimum but simple fusion rules (requiringsimple hardware) that offer reliable and good detection performance.

A better but more complex approach is to possibly identify these compromised SNs and controltheir influence on the FC decision =⇒ Offers an improved detection performance but requiresobserving the SN’s local reports for a period of time. A larger observation time period (K) maylead to a large detection delay that is critical for most of the event detection applications.

We have addressed the fully distributed detection problem and proposed signal processingalgorithms for such an approach =⇒ Very attractive from both the signal processing perspectiveand the communication point of view.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 41 / 42

Page 73: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Key Conclusions

Shown that spatially distributed SNs across the field can offer a reliable operation for eventdetection applications. The system detection performance and the WSN’s operating lifetime canbe further improved by means of resource allocations, optimisation and signal processingalgorithms =⇒ complexity to be kept as simple as possible.

The data fusion problem: we derive the optimal fusion rules (i.e., for attack-free and under-attackWSN scenarios) and have shown that these fusion rules are not implementable in practice andrequire complex local signal processing

=⇒ Derive sub-optimum but simple fusion rules (requiringsimple hardware) that offer reliable and good detection performance.

A better but more complex approach is to possibly identify these compromised SNs and controltheir influence on the FC decision =⇒ Offers an improved detection performance but requiresobserving the SN’s local reports for a period of time. A larger observation time period (K) maylead to a large detection delay that is critical for most of the event detection applications.

We have addressed the fully distributed detection problem and proposed signal processingalgorithms for such an approach =⇒ Very attractive from both the signal processing perspectiveand the communication point of view.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 41 / 42

Page 74: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Key Conclusions

Shown that spatially distributed SNs across the field can offer a reliable operation for eventdetection applications. The system detection performance and the WSN’s operating lifetime canbe further improved by means of resource allocations, optimisation and signal processingalgorithms =⇒ complexity to be kept as simple as possible.

The data fusion problem: we derive the optimal fusion rules (i.e., for attack-free and under-attackWSN scenarios) and have shown that these fusion rules are not implementable in practice andrequire complex local signal processing =⇒ Derive sub-optimum but simple fusion rules (requiringsimple hardware) that offer reliable and good detection performance.

A better but more complex approach is to possibly identify these compromised SNs and controltheir influence on the FC decision =⇒ Offers an improved detection performance but requiresobserving the SN’s local reports for a period of time. A larger observation time period (K) maylead to a large detection delay that is critical for most of the event detection applications.

We have addressed the fully distributed detection problem and proposed signal processingalgorithms for such an approach =⇒ Very attractive from both the signal processing perspectiveand the communication point of view.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 41 / 42

Page 75: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Key Conclusions

Shown that spatially distributed SNs across the field can offer a reliable operation for eventdetection applications. The system detection performance and the WSN’s operating lifetime canbe further improved by means of resource allocations, optimisation and signal processingalgorithms =⇒ complexity to be kept as simple as possible.

The data fusion problem: we derive the optimal fusion rules (i.e., for attack-free and under-attackWSN scenarios) and have shown that these fusion rules are not implementable in practice andrequire complex local signal processing =⇒ Derive sub-optimum but simple fusion rules (requiringsimple hardware) that offer reliable and good detection performance.

A better but more complex approach is to possibly identify these compromised SNs and controltheir influence on the FC decision

=⇒ Offers an improved detection performance but requiresobserving the SN’s local reports for a period of time. A larger observation time period (K) maylead to a large detection delay that is critical for most of the event detection applications.

We have addressed the fully distributed detection problem and proposed signal processingalgorithms for such an approach =⇒ Very attractive from both the signal processing perspectiveand the communication point of view.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 41 / 42

Page 76: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Key Conclusions

Shown that spatially distributed SNs across the field can offer a reliable operation for eventdetection applications. The system detection performance and the WSN’s operating lifetime canbe further improved by means of resource allocations, optimisation and signal processingalgorithms =⇒ complexity to be kept as simple as possible.

The data fusion problem: we derive the optimal fusion rules (i.e., for attack-free and under-attackWSN scenarios) and have shown that these fusion rules are not implementable in practice andrequire complex local signal processing =⇒ Derive sub-optimum but simple fusion rules (requiringsimple hardware) that offer reliable and good detection performance.

A better but more complex approach is to possibly identify these compromised SNs and controltheir influence on the FC decision =⇒ Offers an improved detection performance but requiresobserving the SN’s local reports for a period of time. A larger observation time period (K) maylead to a large detection delay that is critical for most of the event detection applications.

We have addressed the fully distributed detection problem and proposed signal processingalgorithms for such an approach =⇒ Very attractive from both the signal processing perspectiveand the communication point of view.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 41 / 42

Page 77: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Key Conclusions

Shown that spatially distributed SNs across the field can offer a reliable operation for eventdetection applications. The system detection performance and the WSN’s operating lifetime canbe further improved by means of resource allocations, optimisation and signal processingalgorithms =⇒ complexity to be kept as simple as possible.

The data fusion problem: we derive the optimal fusion rules (i.e., for attack-free and under-attackWSN scenarios) and have shown that these fusion rules are not implementable in practice andrequire complex local signal processing =⇒ Derive sub-optimum but simple fusion rules (requiringsimple hardware) that offer reliable and good detection performance.

A better but more complex approach is to possibly identify these compromised SNs and controltheir influence on the FC decision =⇒ Offers an improved detection performance but requiresobserving the SN’s local reports for a period of time. A larger observation time period (K) maylead to a large detection delay that is critical for most of the event detection applications.

We have addressed the fully distributed detection problem and proposed signal processingalgorithms for such an approach

=⇒ Very attractive from both the signal processing perspectiveand the communication point of view.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 41 / 42

Page 78: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Key Conclusions

Shown that spatially distributed SNs across the field can offer a reliable operation for eventdetection applications. The system detection performance and the WSN’s operating lifetime canbe further improved by means of resource allocations, optimisation and signal processingalgorithms =⇒ complexity to be kept as simple as possible.

The data fusion problem: we derive the optimal fusion rules (i.e., for attack-free and under-attackWSN scenarios) and have shown that these fusion rules are not implementable in practice andrequire complex local signal processing =⇒ Derive sub-optimum but simple fusion rules (requiringsimple hardware) that offer reliable and good detection performance.

A better but more complex approach is to possibly identify these compromised SNs and controltheir influence on the FC decision =⇒ Offers an improved detection performance but requiresobserving the SN’s local reports for a period of time. A larger observation time period (K) maylead to a large detection delay that is critical for most of the event detection applications.

We have addressed the fully distributed detection problem and proposed signal processingalgorithms for such an approach =⇒ Very attractive from both the signal processing perspectiveand the communication point of view.

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 41 / 42

Page 79: Distributed Detection and Estimation in Wireless Sensor ...eprints.lincoln.ac.uk/28268/1/Detection and Estimation in...Distributed Detection and Estimation in Wireless Sensor Networks:

Questions/Comments

Edmond Nurellari (University of Leeds) Distributed Detection and Estimation in WSNs June 6, 2017 42 / 42