ni.com

Prototyping LTE-WiFi Interworking on a Single SDR Platform

Walter Nitzold, Clemens Felber, Vincent Kotzsch

National Instruments, Dresden, Germany

• Demand for higher throughput and/or lower latency

• Flexibility within radio access technology (RAT) such as LTE and WiFi

• Transparent End-to-End view involves interworking/coordination

between different RATs

→ How to study/research these interworking technologies?

Future Wireless Communication System (aka 5G and beyond)

eNB

AP

UE + STA

Open Source Protocol Stacks (Layer 2 and above)

IP Networks LTE UE 802.11 LTE EPC

OpenLTELinux 802mac

LTE eNB

5G NR UE

5G NR gNB

Network Simulator NS-3

✓ Open source (GNU GPLv2) discrete-event network simulator in C++

✓ Allows for simulating IP networks including routing algorithms

✓ Provides various wireless/IP simulation models including LTE, Wi-Fi, ...

Only simulation → Integrate SDRs, but how?

NAS

RRC

PDCP

RLC

PHY

MAC

RF + Antenna

Control PlaneUser Plane

Layer 2

(Data Link Layer)

Layer 1

(Physical Layer)

Layer 3

(Network Layer)

Layer 4

(Transport Layer)

Air transmission through EM waves

Channel mux, adaptive resource scheduling (incl

Beamforming), HARQ, access procedure,

carrier aggregation

Coding, modulation, resource mapping, MIMO

Segmentation, reassembly, ARQ

Integrity, ciphering, duplicate detect, SN,

reordering, dual connectivity

SDR Integration

IP

??? ???L1-L2 API L1-L2 API

OSI Model 3GPP LTE/5G Protocol Stack

L1-L2 API

RF HW

802.11

MAC low + PHY

802.11 MAC high

functionality

3rd party

802.11

protocol

stack

WiFi

Aligned on

concepts,

mechanisms,

general structure,

…

Cellular

L1-L2 API

RF HW

LTE PHY/L1

< 6GHz

5G/NR PHY/L1

> 6GHz

RF HW

LabVIEW based

MAC/L2

L2-L3 API

3rd party L3

3rd party L2

3rd party

protocol

stack

(e.g. ns-3)

MAC-PHY API Concept

Potential of a general L1-L2 API

▪ Offers the same ease of use and mechanisms for different physical layer

implementations.

▪ Enables very flexible configuration and E2E user data transmission.

▪ Enables faster adoption, extension, migration and integration of

physical layer prototyping systems towards protocol and network level

research & applications.

▪ Increases the level of system abstraction, because of no need to understand

all details of the underlying physical layer.

▪ Offers a high level of re-use and increases the transparency of a complex

wireless system.

ni.com

How does the L1-L2 API work?

L1-L2 API – General principles

Communication based on a common set of 3 message types

PHY MAC

REQ → (Service) Request

IND → Indication

(status, data, error, …)

CNF → Confirm

w/ option to be disabled

PAYLOAD

CONTROL

Example: TX REQ for data channel

MCS, RB allocation,

TB size, …

Transport block

bits/bytes

L1-L2 API in action - Message Sequence Chart

PHY_UL_TX_CONFIG_REQ(SFN, slot index n+2, ULSCH_TX_CONFIG(PRB alloc, MCS, RNTI, PDU Idx)

PHY_UL_TX_PAYLOAD_REQ(SFN, slot index n+2,ULSCH_TX_MAC_PDU, PDU Idx)

LTE UE L1 LTE UE L2

PHY_TIMING_IND(SFN, slot index n+1, …)

PHY_ULSCH_ASSIGNMENT_IND([SFN, slot index n+1,]TB size, HARQ PID, NDI, [MCS, PRB alloc] …)

LTE eNB L1(Timing Master)

LTE eNB L2

PHY_TIMING_IND(SFN, slot index n, …)

PHY_DL_TX_CONFIG_REQ(SFN, slot index n+1, DCI_TX_CONFIG(DL_Grant, UL_grant)

DLSCH_TX_CONFIG(PRB alloc, MCS, RNTI, PDU Idx))

PHY_DL_TX_PAYLOAD_REQ(SFN, slot index n+1,DLSCH_TX_MAC_PDU(PDU Idx, MAC PDU))

PDSCH/PDCCH & PUSCH/PUCCH (slot index n)

PHY_DLSCH_RX_IND(SFN, slot index n+1,DLSCH_RTX_MAC_PDU)

PHY_UL_RX_CONFIG_REQ(SFN, slot index n+1,ULSCH_RX_CONFIG [0])

1ms

PHY_TIMING_IND(SFN, slot index n+1, …) PDSCH/PDCCH & PUSCH/PUCCH (slot index n+1)

PDSCH/PDCCH & PUSCH/PUCCH (slot index n+2)

PHY_TIMING_IND(SFN, slot index n+2, …)

PHY_DL_TX_CONFIG_REQ(SFN, slot index n+2, DLSCH_TX_CONFIG(PRB alloc, MCS, RNTI, PDU Idx)

DCI_TX_CONFIG(DL_Grant, UL_grant)

PHY_DL_TX_PAYLOAD_REQ(SFN, slot index n+2,DLSCH_TX_MAC_PDU, PDU Idx)

PHY_UL_RX_CONFIG_REQ(SFN, slot index n+2,ULSCH_RX_CONFIG [0])

PHY_ULSCH_RX_IND(SFN, slot index n,ULSCH_TX_MAC_PDU)

PHY_ULSCH_RX_IND(SFN, slot index n+1,ULSCH_TX_MAC_PDU)

PHY_TIMING_IND(SFN, slot index n+2, …)

PHY_TIMING_IND(SFN, slot index n, …)

PHY_UL_TX_CONFIG_REQ(SFN, slot index n+1, ULSCH_TX_CONFIG(PRB alloc, MCS, RNTI, PDU Idx)

PHY_UL_TX_PAYLOAD_REQ(SFN, slot index n+1,ULSCH_TX_MAC_PDU, PDU Idx)

PHY_ULSCH_ASSIGNMENT_IND(SFN, slot index n,PRB alloc, MCS, RNTI)

PHY_DLSCH_RX_IND(SFN, slot index n,DLSCH_RTX_MAC_PDU)

PHY Timing IND

Transport

medium

btw.

L1-L2

L1-L2 API – Interfaces and Modules

L2

INDs

REQs

L1 implementation

specific interface

PHY mode control

CNFs

Msg

GenMsg

Handler

Msg

Handler

L1

Con

tro

l

L1

L2 implementation

specific interface

Implementation specific, e.g.

- P2P FIFOs

- H2T/T2H FIFOs

- Memory pointers

- C-API

- UDP via Ethernet

- …

„write“

„read“

„read“

„write“Msg

Gen

Can be realized by standard

platform components

REQs

PHY Timing IND

INDs

CNFs

Common message interface

To be provided by L1

compliant with common API

To be provided by L2

compliant with common API

RRC

LTE-WiFi Prototyping Setup with USRP-2974

LV RT LTE API

Handler

LV

RT

WiF

i AP

I

Han

dle

r

LV FPGA LTE AFW

DAC/ADC & RF

PCIe Expansion

USRP 2953

USRP 2974

Inte

l i7

CP

UX

ilin

xK

7

FP

GA

LV FPGA 802.11 AFW

DAC/ADC & RF

Xili

nx

K7

FP

GA

DMA DMA

Pipes

UD

P

Ns-3 simulation:

• Internal client /

serv or TapBridge

• Internet

• LTE UE or eNB

• WiFi AP or STA

eNB

AP

UE + STA

Measurement of End-to-End Throughput

eNB

AP

UE + STA

• Close to capacity limit of LTE PHY layer

• Throughput Increase compared to pure software-based PHY

Conclusions and Outlook

▪ Generalized API specification to connect PHY and MAC of different RATs

▪ Interworking research is possible with ns-3 and attached real-time SDR

▪ Implementation is used in EU-funded research project ORCA (https://www.orca-project.eu)

▪ Enhanced ns-3 with L1-L2 API implementation available under https://github.com/ni/NI-ns3-ApplicationExample

Next steps:

▪ Maintain and further extend based on community/research needs

▪ Investigation and application to other higher layer stacks like Open Air Interface

▪ Investigation and application to other physical layers like 5G GFDM PHY by TUD

ni.com

Thanks!