View
215
Download
0
Category
Preview:
Citation preview
8/15/2019 Energy Efficient MAC Protocol for Multimedia IoT devices
1/6
Adaptive Duty Cycling based Multi-hop PSMP for
Internet of Multimedia Things
Bilal Afzal, Sheeraz A. Alvi, Ghalib A. ShahAl-Khawarizmi Institute of Computer Science, UET, Lahore 54000, Pakistan
Email contact: {bilal.afzal, sheeraz.akhtar, ghalib}@kics.edu.pk
Abstract—In several use-cases of Internet of Things (IoT),IEEE 802.11 based WLANs are more favorable due to superiordata rate support even though their energy efficiency is notup to the mark. Particularly, wireless multimedia sensors basedWLANs demand higher energy resources. In this regard, variousIEEE 802.11 based power saving mechanisms are developed.IEEE 802.11n standard specifies Power Save Multiple Poll(PSMP) protocol. However, PSMP is infeasible for many IoTbased systems specifically in use-cases where multi-hop com-munication is required. Moreover, PSMP scheduling mechanismlacks the capability to adapt to the dynamic Quality of Service
(QoS) requirements in Internet of Multimedia Things (IoMT). Inthis paper, a QoS aware Multi-Hop PSMP (mPSMP) protocol isproposed to enable energy efficient multimedia communicationover IoT. The mPSMP incorporate a traffic scheduling modelto allocate channel resources in a time division multiple accessmanner. Therein, adaptive duty cycling is employed to minimizeenergy utilization, while assuring the required multimedia QoSfor each node. The proposed protocol is implemented in NetworkSimulator-2 (NS-2). Analytical analysis and simulation study sug-gests reduction in end-to-end delay and duty cycling along withsignificant improvement in energy efficiency of IoMT devices.
Index Terms—Internet of Things, Multimedia sensors, PSMP,Energy efficiency, Multi-hop communication.
I. INTRODUCTION
Internet of Things (IoT) is characterized as the notion of inter-connected ‘things’ which are uniquely identifiable and
able to communicate with any device connected to the Internet
[1]. Recent escalation in the utilization of multimedia services
and applications such as video conferencing, telemedicine,
online-gaming, etc., incited an ostentatious growth of the band-
width hungry multimedia content. Moreover, the availabil-
ity of Complementary Metal Oxide Semiconductor (CMOS)
cameras and microphones has attracted lot of research on
the Wireless Multimedia Sensor Networks (WMSNs) [2],
[3], wherein resource constrained devices retrieve multimedia
content from the physical environment.
The ‘Internet of Multimedia Things’ (IoMT) is a novel
paradigm whose prime objective is to enable multimediastreaming as part of the realization of IoT. The IoMT
paradigm enables a wide range of applications in the areas of
home/building security, smart cities, traffic monitoring, and en-
ergy management [4]. The wireless communication technology
proposed for IoT systems, i.e. ZigBee, is designed for sensor
network application requiring limited data rate of 250 Kbps.
This data rate is infeasible for multimedia applications and
particularly for real-time multimedia communication. There-
fore, IEEE 802.11 standard (Wi-Fi) is a potential alternative
for IoMT and has already been adopted for WMSNs due to
its higher data rate support [5], [6]. Nevertheless, the current
power saving mechanisms in IEEE 802.11 standard, i.e. Power
Save Mechanism (PSM), Power Save Multiple Poll (PSMP),
lack the capability to adapt to the dynamic Quality of Service
(QoS) requirements of IoMT systems.
In [7], it is suggested that the transmit power of energy con-
strained devices along multi-hop routes in dense wireless net-
works to be kept limited in order to conserve energy and avoid
interference. Consequently, power utilization is improved withthe increase in number of hops between source and destination
nodes. Likewise, IoT devices are mostly battery operated,
posses short range radios, and hence have limited transmis-
sion power capabilities. Thus, Low-power IoT devices may
benefit from multi-hop communication mechanism to reduce
the energy consumption. Existing IEEE 802.11 power saving
standards, however, only support single-hop communication.
Therefore, IEEE 802.11 power saving mechanisms need to be
optimized to enable energy efficient multi-hop communication
over IoMT while meeting the desired QoS requirements.
In this paper, an adaptive duty cycling based multi-hop
PSMP protocol is proposed which enables energy efficient
multi-hop communication in infrastructure based WLANs.The proposed mPSMP protocol extends the single-hop PSMP
protocol specified by IEEE 802.11n standard. The QoS re-
quirements of the multimedia traffic, particularly real-time
traffic, is considered by incorporating a traffic scheduling
model. The model allocates traffic opportunity to nodes in
a Time Division Multiple Access (TDMA) fashion to fulfill
the delay bound of packets per second in order to realize
multimedia communication over IoMT. Moreover, the energy
utilization and end-to-end delay of each node is minimized by
employing adaptive duty cycling at each network node which
then aggregate their generated packets while considering the
delay restriction to ensure required QoS.
The main contributions of this paper are as follows:
• A multi-hop communication protocol is proposed for
multimedia communication over IoMT.
• A QoS aware bandwidth allocation algorithm is designed
to support real-time multimedia traffic.
• Adaptive duty cycling is incorporated and a traffic
scheduling model is designed for performance analysis.
• Frame aggregation is utilized to minimize protocol over-
head and conserve energy consumption.
8/15/2019 Energy Efficient MAC Protocol for Multimedia IoT devices
2/6
I I . RELATED W OR K
The design of low-power hardware modules and recent work
in literature focused on reducing energy consumption along
with improving energy efficiency of IEEE 802.11 standard
made it a good candidate for IoT [6], [8], [9]. In [9], feasibility
of low-power Wi-Fi sensors for IoT is studied and energy
consumption results in different power states are compared
with 6LowPAN. Owing to higher data rates, low-power Wi-Firadios are proved to perform better when bigger data packets
are communicated, in multimedia streaming for example, since
packet fragmentation is avoided.
IEEE 802.11 standard based power saving schemes pro-
posed in [10], [11], and [12] curtail energy consumption by
regulating the number of nodes in awake mode and schedul-
ing beacons aperiodically to minimize duty cycling. In [12],
centralized PSM (C-PSM) is proposed in which Access Point
(AP) selects optimal values of beacon interval and reported
energy efficiency as much as 76% compared to standard PSM.
However, C-PSM lacks support for multi-hop communication.
Higher number of nodes in a network results in increased
interference and contention time; thus nodes stay awake for
longer durations. To mitigate these issues, authors in [13] sug-
gested to divide beacon interval into time slots and assigning
them to individual nodes in a TDMA like mechanism, allowing
them to only wake up at their scheduled time slots. In [14],
authors proposed a scheduling mechanism named NAPman
which controls the traffic destined for PSM-enabled stations
so that the other stations in the network do not starve.
The authors in [15] has proposed a Congestion Aware-
Delayed Frame Aggregation (CA-DFA) algorithm, wherein the
transmissions are intentionally delayed and only transmitted
utilizing frame aggregation when the congestion level drops to
a certain threshold. Similarly, another buffering scheme namedLow Energy Data-packet Aggregation Scheme (LEDAS) is
proposed in [16]. However, the LEDAS scheme does not
consider application specific delays induced by buffering
particularly for real-time multimedia traffic. These proposed
schemes lacks the support for multi-hop communication.
Recently, a multi-hop IEEE 802.11 PSM mechanism named
as MH-PSM is presented in [17] for multi-hop toy-to-toy
communication. The proposed traffic announcement scheme
enables nodes along multi-hop route to stay awake only if
there is pending traffic for them and hence, reducing significant
associated cost with mandatory wake-ups at each beacon
intervals. However, in MH-PSM when multiple nodes contend
for the channel to transmit PS-Poll frames to retrieve theirpackets, the collision probability increases which results in
energy and bandwidth wastage.
In PSMP, AP schedule uplink (UL) and downlink (DL)
transmissions of nodes in a TDMA manner, and afterwards
they can go into sleep mode. Hence, energy and bandwidth
overhead is significantly reduced compared to other power
saving mechanisms. However, PSMP is infeasible for multi-
hop IoT scenarios and the scheduling mechanism does not
adapt to dynamic QoS requirements of IoMT systems.
III. PROPOSED M ULTI-H OP PSMP PROTOCOL
A. Problem Definition
Real-time multimedia communication in IoMT environment
necessitate stringent QoS traffic requirements in terms of
bandwidth, delay, jitter, and reliability. In addition, the frame
aggregation threshold, already proven to be bandwidth and
energy efficient in prior research studies [15], is another
critical network metric. Moreover, frame aggregation is inturn dependent upon supported data rate and application
specific delay bounds corresponding to various multimedia
devices. Therefore, multimedia communication over multi-hop
networks requires an efficient energy saving algorithm which
is aware of the dynamic QoS requirements and also take into
account various interdependent parameters such as data rate,
frame rate, frame aggregation threshold and delay bounds
while reducing duty cycling of multimedia network devices.
Considering these unique compulsions, we formulate an ILP
problem whose objective function is to minimize duty cycling
utility function, U u(∂ i, γ i, Ψi), subject to the delay bound∂ i, supported data rate γ i and frame aggregation threshold
Ψi constraints. Let N u = {u1, u2,.....,un} be the set of multi-hop nodes ui and let Qu,i = { p1, p2, . . ,pm}, be theircorresponding set of queues. The problem is formulated as:
Minimize:
ui∈N u
U u(∂ i,γ i,ψi) ∀ui ∈ N u (1)
Subject to:
m
i=1
∂ p,i ≤ ∂ i,max ∀ p ∈ Qu, ∀ui ∈ N u (2)
ρ p,iψu,i
γ u,i ≤ ∂ p,1 ∀ p ∈ Qu,∀ui ∈ N u (3)
n
i=1
nτ u,i ≤ 1 ∀ui ∈ N u (4)
The Constraint (2) in above formulation ensures that the delay
associated with any packet pi buffered at the queue Qu of
station ui, should not exceed the maximum delay bound limit,
∂ i,max. In addition, the delay induced by buffering a packet
in queue is in turn dependent upon the frame aggregation
threshold and data rate; thus Constraint (3) mandates that
aggregated packets are transmitted within the limitation of
delay bound of oldest packet in queue, (∂ p,1). While, the
Constraint (4) corresponds to the allocation of TransmissionOpportunity (TXOP) τ u,i, and ensures that QoS requirements
specified by frame rate are met within one second. In real-
time multimedia streaming, the variation in QoS requirement
for every multi-hop away node is variable and hence affects
the duty cycling of a node as well as its child-nodes. Thus,
the solution to the given ILP problem is found out to be NP-
Complete and computationally expensive in terms of energy
requirement. However, the multimedia devices in an IoMT sys-
tem are inherently resource constrained for such computations.
8/15/2019 Energy Efficient MAC Protocol for Multimedia IoT devices
3/6
(a) Time-line of Multi-hop PSMP0 00 00 01 11 11 1
Internet of Things
STA 1 STA 3
STA 5
STA 4
STA 2
AP
(b) Topology
Fig. 1: mPSMP Operation for a 5 Node Network
In this section, we provide the operational details of the
proposed multi-hop PSMP protocol, referred to as ‘mPSMP’.The mPSMP protocol enables multi-hop operations in PSMP
based WLANs while minimizing the utility function of duty
cycling. The protocol adaptively selects appropriate frame
aggregation threshold and the TXOP duration for each node
based on its uplink schedule time, data transmission rate, the
QoS specified minimum frame per seconds and application
specific delay bound with respect to the arrival time of the
oldest pending packet in queue. When a data packet is received
at the MAC layer, it is appended in the pending packets queue
maintained at each node. Based on Channel State Information
(CSI) received from the intended receiver node, the transmitter
node determines the highest supported data rate for the given
channel conditions. Higher data rate alleviates the need of longer PSMP Uplink Transmission Time (UTT) and Downlink
Transmission Time (DTT) duration requirement, thus band-
width resources are conserved. The aggregated packets are
then transmitted as soon as the amount of data in the queue
exceeds the frame aggregation threshold or the delay for oldest
packet equals the maximum delay bound limit.
Once a transmitting node determines the aggregation thresh-
old and achievable data rate, it can ask AP (or parent node) for
the TXOP or more accurately PSMP-UTT service period. In
a PSMP sequence, AP (or parent node) shares its own TXOP
to provide PSMP enabled nodes to transmit uplink traffic
and/or receive downlink traffic. Moreover, as highlighted in
[16], another pertinent issue is the requirement for a node tostay awake for its allocated TXOP even if it has no more
traffic to send in current PSMP sequence. This results in
energy and bandwidth under utilization. This issue is resolved
in mPSMP protocol by incorporating adaptive duty cycling,
i.e. if a node has no more packets to be sent in its allocated
TXOP, it can sleep for the rest of its PSMP-UTT duration,
which helps in minimizing the duty cycling utility function.
Lastly, the traffic demand and associated parameters of delay
bound, frame aggregation, frame rate and data rate are being
fed to the traffic scheduling model explained in next section.
B. mPSMP Protocol Operation
The operation of proposed mPSMP protocol is described as
follows; and a sample operation for a five node network topol-
ogy Fig. 1b is shown in Fig. 1a. For detailed understanding
of PSMP operation, reader is referred to [18].
• Firstly, AP advertise its service set identifier (SSID) peri-
odically using Beacon frame which has Timestamp sub-
type containing value of stations synchronization timer at
the time the frame was transmitted.
• This enables synchronization between AP and stations
and accordingly, AP assign them association IDs (AIDs)
starting from AID 1. After this, AP send PSMP Action
frame, indicating if there is any traffic buffered for nodesin Traffic Indication Map (TIM) field.
• The PSMP frame’s STA Info field notifies the TXOP for
each node including their PSMP-DTT and PSMP-UTT
Duration, and PSMP-DTT and PSMP-UTT Start Offset.
• Correspondingly, each one-hop node knows what state it
should be in at particular times. The schedule is assigned
in the order of Association IDs (AIDs) of one-hop away
nodes, i.e. STA 1’s schedule leads STA 2’s, and so on.
Hence, only a single node stays awake for the time
duration specified for it in the STA Info field.
• In this way, each one-hop away node wakes up as per the
order given in TIM bit of PSMP Action frame, receives
its DL transmission from AP in PSMP-DTT and upon itscompletion fell asleep for the remaining duration of DL
traffic transmission of other nodes.
• The PSMP-DTT is followed by PSMP-UTT, therein each
one-hop away node wakes up on its turn and sends
its buffered UL traffic towards AP and then fell asleep
when its PSMP-UTT duration expires. In the first PSMP
sequence, all one-hop away nodes are served by AP.
• In subsequent PSMP sequences, two-hop away nodes get
their traffic schedule by their respective one-hop parents.
8/15/2019 Energy Efficient MAC Protocol for Multimedia IoT devices
4/6
The parent nodes initiate PSMP sequences based on their
AIDs, i.e. PSMP sequence initiated by STA 1 leads that
of STA 2, and so on. Similarly, each one-hop away node
will act as an AP for its child-nodes.
• To specify a single wake up interval for child-nodes,
all PSMP sequences are set apart to the maximum
PSMP sequence duration specified in standard [18], i.e.
8.184msecs. Therefore, even if the previous PSMP se-
quence DTT and UTT durations change, still the PSMP
sequence interval of others is not disturbed. This enables
efficient duty cycling, as shown in Fig. 1a.
• The parent nodes need to wake up for the PSMP-frame
and their PSMP-DTT and PSMP-UTT durations. In ad-
dition, they have to stay awake in the PSMP sequence
which they are providing to their child-nodes. Rest of
the time they can sleep to conserve energy.
• Likewise, the child-nodes which are not supporting other
nodes, are required to receive the PSMP frame from their
respective parents and only stay awake for their PSMP-
DTT and PSMP-UTT durations.
•
In the subsequent PSMP sequence durations, the samepattern is repeated iteratively for three hop away nodes,
and so on, until the traffic demand of all the nodes is
served. After these PSMP sequences, the AP starts the
contention period till the next phase of PSMP sequences
as specified in standard [18].
Although, the algorithm outlined above is implemented and
optimized considering an infrastructure based network sce-
nario, however, the mPSMP protocol can also be adapted in
an ad hoc network scenario. The only major difference will
be that in current implementation, the AP is kept awake all
the time as it is considered to have a power line. While in
ad hoc mode, the node which starts the first PSMP sequence
will go into sleep state upon completion of its PSMP sequenceduration; and afterwards the child-nodes of next hop will be
served in subsequent PSMP sequence durations.
IV. PERFORMANCE A NALYSIS
In this Section, a traffic scheduling model is presented con-
sidering the QoS traffic requirements of IoMT based systems.
Let the time spent in transmission is denoted by T tx and power
consumed in transmission state by P tx. Similarly, let the time
spent in sleep mode is denoted by T sl and power consumed by
node in sleep state by P sl. Let λ represents frame rate that is
number of frames to be transmitted by a node in one second in
order to support the QoS requirements. Ensuring frame rate is
important to support smooth streaming of multimedia content.It is essential to be considered in order to keep the jitter level
within some pre-defined bound to provide satisfactory Quality
of Experience (QoE).
For each frame, ψ number of packets are aggregated based
on the achievable data rate and QoS requirement of a node. As
specified earlier, it is critical to keep the delay bound of the
queued packets in consideration while adaptively selecting the
frame aggregation at each node. Thus, the number of packets
needs to be sent by a node in one second are λ × ψ. Let the
size of a single packet is ρ bits. Correspondingly, required
amount of throughput Γ per second can be given as:
Γ = λ × ψ × ρ bits per sec (5)
Similarly, given the data rate of γ Mbps, the time τ required
by each node to transmit this data while satisfying the QoS
requirement of λ frames per second can be given as:
τ = Γ
γ = λ × ψ × ρ
γ secs (6)
Thus, τ is the cumulative PSMP UTT duration required
to send Γ amount of data by a single node in one second inorder to satisfy the delay bound. However, since the maximum
duration of PSMP sequence is 8.184 msecs, therefore this
cumulative time is scheduled in multiple PSMP sequences.
Within a single PSMP sequence, the Downlink transmission
time (i.e. DLt) and Uplink transmission time (i.e. U Lt) are
allocated depending upon the number of child-nodes to be
scheduled in single PSMP-sequence belonging to each parent
node, at a specific hop level. If there are number of nodes
required to be scheduled in single PSMP sequence, then
the possible per node allocated transmission time for bothDownlink and Uplink traffic can be calculated as:
DLt + U Lt = 8.184
msecs (7)
Considering PSMP-DTT and PSMP-UTT durations to be
equal, DLt = U Lt = T t, implies
2 × T t = DLt + U Lt (8)
T t = 8.184
2 × msecs (9)
Here T t is the transmission time.In a single-hop scenario, back-to-back PSMP sequences
can be initiated (PSMP bursts). However, to enable multi-hop
communication and to provide access to nodes located outsidethe transmission range of AP, each single-hop away node needs
to carry its child-nodes traffic towards AP along with its own
traffic. Hence, assuming the traffic demand of every child-
node is same as their peers belonging to other single-hop away
node; if every i node has j number of child-nodes, then after
determining the required amount of transmission time for node
i i.e. T t,i, the updated required number of PSMP sequences Λfor the node i within one second can be calculated as:
Λi,j = τ × j
T t,i, 0 ≤ i, j ≤ (10)
This essentially means that multiple PSMP sequences are
scheduled in one second to facilitate the traffic demand of every child-node. Moreover, the time after which a specific
PSMP sequence is scheduled again to allocate TXOP for the
child-nodes is referred as PSMP interval. Based on the value of
Λi,j , we can now determine the value of PSMP interval δ , afterwhich TXOP is assigned again to a node (this is uniformly
distributed over one second):
δ = 1
j=1
Λi−1,j
(11)
8/15/2019 Energy Efficient MAC Protocol for Multimedia IoT devices
5/6
Fig. 2: Energy consumption: TDMA vs mPSMP
Therefore, T t can now be calculated by adding the total
TXOPs allocated to all the nodes within one second. Using
equation 9 and 10 we calculate T aw,i, which is the total time
a node i stays in awake state in one second duration:
T aw,i =
i=1,j=1
Λi,j(2××T t,i+3×T sifs+T pf +T wait) (12)
Here T sifs represents the number of SIFS intervals in each
PSMP sequence, T wait is the additional short time spend for
each node to stay awake before sending a lost TXOP request
towards AP, and T pf is the time spent in sending PSMP
frames. Likewise, the total sleep time of a node i denoted
by T sl,i can be given as:
T sl,i = 1 − T aw,i (13)
Each node i fell sleep between the time duration of any of
its two TXOPs as described in proposed algorithm. However,
if a node is also acting as the parent of any next hop child-
node, then it has to stay awake till the time duration itcommunicate with and exchange frames to its child-nodes.
The entire methodology adapted for this mathematical model
works in a TDMA like fashion while fulfilling the traffic
demand of each node. Finally, the energy consumed by a node
in various states can be calculated as follows:
E = T sl × P sl + T aw × P tx (14)
E sl = T sl × P sl = (1 − T aw) × P sl (15)
E energysaving = E sl
E (16)
EnergyEfficiency(%) = E total − E
E total× 100% (17)
Here E , E sl, and E total, represent the energy consumption
in active state, in sleep state and the total energy consumed,
respectively. Let E represents the energy consumption without
applying mPSMP protocol, then the energy efficiency can be
given as:
η = E
E (18)
Fig. 3: Energy efficiency: PSMP vs mPSMP
V. PERFORMANCE E VALUATION
In this Section, we present the simulation model to evaluate
the performance of mPSMP protocol. The simulations consist
of an infrastructure based IoT network Fig. 1b, with an AP
and 5 stations (STAs) each generating constant bit rate (CBR)
traffic. The carrier sensing range of STAs is set such that the
connectivity between any two stations is ensured. We vary
the packet size from 128 to 2048 bytes and packet interval
from 0.01 to 0.1 secs with uniform distribution. Moreover,
the following values are considered for the energy model; Tx
Power = 660mW, Rx Power = 395mW, Idle Power = 35mW,
Sleep Power = 1mW, Initial Energy = 1000J.
Simulation results are averaged for multiple flows of several
distinct topologies and each simulation runs for 100 secs
duration. Unless specified, we used the packet size of 512
bytes, packet interval of 0.05 secs and frame rate of 25 frames
per second. We investigate the aggregate energy consumption,
by keeping the packet interval fixed at 0.05 secs while varying
the packet size from 128 to 2048 Bytes. The simulation model
also validates the traffic scheduling model of Section IV.The analytical results computed from traffic scheduling
model are compared with that of simulation results are shown
in Fig. 2. Evidently, the increase in packet size results in more
energy consumption due to the effective increase in throughput
at each station as more time is consumed in awake state to
meet the QoS requirement of 25 frames per second. However,
as depicted in Fig. 3, mPSMP significantly elevates the aggre-
gate energy efficiency (24% on average) and the increase in
packet size has only marginal effect on the energy efficiency
(4% cut) of multi-hop stations employing our mPSMP protocol
compared to the detrimental effect (20% cut) of packet size
variation (increase) in the absence of mPSMP mechanism.
Similarly, Fig. 2 also provides a comparison of mPSMP withthe traditional TDMA approach of IEEE 802.11 standard
without duty cycling and shows significant performance gains
in terms of energy consumption by employing mPSMP.
In standard PSMP protocol a station is not the TXOP holder
which essentially means that it stays awake for its allocated
TXOP even if there are no packets buffered in its queue.
mPSMP employs adaptive duty cycling which enables a station
to sleep dynamically after sending its pending traffic. The
impact of adaptive duty cycling on energy consumption for
8/15/2019 Energy Efficient MAC Protocol for Multimedia IoT devices
6/6
Fig. 4: Energy efficiency gain due to adaptive duty cycling (*)
multi-hop stations across the flow (STA-5→STA-4→STA-2)can be visualized in Fig. 4. This impact is enhanced with
increase in packet interval particularly for stations along multi-
hop route from AP. Thus, additional energy is conserved which
indicates the benefits of adaptive duty cycling in reducing
idle listening time, especially for multi-hop network topologies
where traffic generation rate varies over the period of time.
The average end-to-end delay results are shown in Fig. 4.As expected, end-to-end delay increases with an increase in
packet size. It is incremented proportionally with the increase
in number of hops; however thanks to the mPSMP protocol,
the maximum average end-to-end delay for 3 hop away nodes
is 20ms which is very much lower compared to the vari-
ous implementations of existing power saving mechanisms.
Moreover, while some packets are delayed and forwarded over
multi-hops in more than one PSMP Sequence; but QoS will be
guaranteed by the traffic scheduling model according to given
frame rate. It also ensures that the end-to-end delay for frames
at a given station is kept within per second delay bound.
VI . CONCLUSION AND F UTURE W OR K
The existing IEEE 802.11 based energy saving mechanisms
do not meet the multi-hop communication and dynamic QoS
requirements. In this paper, mPSMP protocol is proposed
to enable multi-hop communication in IEEE 802.11 IoMT
systems. Moreover, a traffic scheduling model incorporating
adaptive duty cycling is designed to meet the QoS require-
ments of resource constrained multimedia devices. Simulation
results indicate reduction in duty cycling and end-to-end delay
along with significant improvement in energy efficiency of
multi-hop nodes. In future, we aspire to make our protocol
more adaptive by relaxing the assumption of uniform traffic
distribution for stations by separately selecting QoS parameters
in accordance to distinct traffic classes; and in addition, to meet
the unique QoS specifications for various use-cases of IoMT.
ACKNOWLEDGEMENT
This work is supported by the National ICT R&D Fund,
Gov. of Pakistan under Grant no. ICTRDF/TR&D/2013/04.
REFERENCES
[1] Luigi Atzori, Antonio Iera, and Giacomo Morabito. The internet of things: A survey. Computer networks, 54(15):2787–2805, 2010.
Fig. 5: Data rate effect on traffic delay
[2] Mohammad Ashraful Hoque, Matti Siekkinen, and Jukka K Nurminen.Energy efficient multimedia streaming to mobile devicesa survey. Com-munications Surveys & Tutorials, IEEE , 16(1):579–597, 2014.
[3] Ian F Akyildiz, Tommaso Melodia, and Kaushik R Chowdhury. Asurvey on wireless multimedia sensor networks. Computer networks,51(4):921–960, 2007.
[4] Jayavardhana Gubbi, Rajkumar Buyya, and Slaven et al. Marusic.Internet of things: A vision, architectural elements, and future directions.Future Generation Computer Systems, 29(7):1645–1660, 2013.
[5] Low Energy Consumption for Networks of Future. Available Online,http://www.econet-project.eu/, 2011.
[6] Serbulent Tozlu, Murat Senel, Wei Mao, and Abtin Keshavarzian.Wi-fi enabled sensors for internet of things: A practical approach.Communications Magazine, IEEE , 50(6):134–143, 2012.
[7] Mira Andriani, M Irhamsyah, et al. Analysis of energy efficiency forwi-fi 802.11 b multi-hop networks. In Communication, Networks and Satellite (COMNETSAT), 2013 IEEE International Conference on, pages64–68. IEEE, 2013.
[8] MICROCHIP. ww1.microchip.com/downloads/en/devicedoc/70005171a.pdf[9] Serbulent Tozlu. Feasibility of wi-fi enabled sensors for internet of
things. In Wireless Communications and Mobile Computing Conference(IWCMC), 2011 7th International, pages 291–296. IEEE, 2011.
[10] Justin Manweiler and Romit Roy Choudhury. Avoiding the rush hours:Wifi energy management via traffic isolation. In Proceedings of the 9thinternational conference on Mobile systems, applications, and services ,pages 253–266. ACM, 2011.
[11] Daewon Jung, Ryangsoo Kim, and Hyuk Lim. Power-saving strategyfor balancing energy and delay performance in wlans. Computer Communications , 50:3–9, 2014.
[12] Yi Xie, Xiapu Luo, and Rocky KC Chang. Centralized psm: an ap-centric power saving mode for 802.11 infrastructure networks. In Sarnoff Symposium, 2009. SARNOFF’09. IEEE , pages 1–5. IEEE, 2009.
[13] Pavlos Petoumenos, Georgia Psychou, and Kaxiras et al. Mlp-awareinstruction queue resizing: The key to power-efficient performance. In
Architecture of Computing Systems-ARCS , pages 113–125. Springer,2010.
[14] Eric Rozner, Vishnu Navda, Ramachandran Ramjee, and ShravanRayanchu. Napman: network-assisted power management for wifidevices. In Proceedings of the 8th international conference on Mobilesystems, applications, and services, pages 91–106. ACM, 2010.
[15] Daniel Camps-Mur, Manil Dev Gomony, and Prez-Costa et al. Leverag-ing 802.11 n frame aggregation to enhance qos and power consumptionin wi-fi networks. Computer Networks, 56(12):2896–2911, 2012.
[16] Rajesh Palit, Kshirasagar Naik, and Ajit Singh. Impact of packetaggregation on energy consumption in smartphones. In Wireless Com-munications and Mobile Computing Conference (IWCMC), 2011 7th
International, pages 589–594. IEEE, 2011.[17] Ioannis Glaropoulos, Stefan Mangold, and Vladimir Vukadinovic. En-
hanced ieee 802.11 power saving for multi-hop toy-to-toy communi-cation. In Green Computing and Communications (GreenCom), 2013,pages 603–610. IEEE, 2013.
[18] IEEE 802.11n-2009 Standard for Information technology, Local and metropolitan area networks, Specific requirements, Part 11: Wireless
LAN Medium Access Control (MAC)and Physical Layer (PHY) Specifi-cations Amendment 5: Enhancements for Higher Throughput . IEEE.
Recommended