The Nucleus Approach: Efficient Development of

Portable Wireless Communication Transceivers

T. Kempf, J. Castrillon, G. Ascheid, R. Leupers, H. Meyr

Institute for Communication Technologies and Embedded Systems,

RWTH Aachen University, Germany

Torsten.Kempf@ice.rwth-aachen.de

February 21, 2011

Software Defined Radios (SDRs) are an attractive solution to cope with theflexibility of mobile wireless communication devices in future. The main idea isto implement functions in software that have been previously integrated as hard-wired logic circuits. Initial results of available SDR platforms highlight the largepotential, but also their limitations. In particular, the limited energy budget formobile telecommunication devices creates a huge barrier for SDR approaches,which need to be addressed carefully by specific design optimizations. Anotherlimitation is that software development has to deal with low-level hardware de-tails in order to achieve the required performance. Unfortunately, this limitsthe portability of radio standards as the software becomes platform dependent.Addressing this issue, the Nucleus methodology [1, 2] has been created thatcovers the key issues of SDR development, namely novel algorithms, hardwarearchitectures and tooling.

In this poster we present the Nucleus methodology and the related tooling.Furthermore we present how it can be applied to a MIMO OFDM transceiverthat has been developed under grant FP7-248716-2PARMA.

References

[1] Ramakrishnan, V., Witte, E. M., Kempf, T., Kammler, D., Ascheid, G. and H.Meyr, RWTH Aachen University; Adrat, M. and M. Antweiler, Dept. FKIE/KOM,Wachtberg, Germany. Efficient and Portable SDR Waveform Development: TheNucleus Concept. In IEEE Military Communications Conference (MILCOM 2009),Boston, USA, Oct 2009.

[2] Castrillon, J., Schuermans, S., Stulova, A., Sheng, W., Kempf, T., Ishaque, A.,Leupers, R., Ascheid, G. and H. Meyr. Component-based Waveform Development:the Nucleus Tool Flow for Efficient and Portable SDR. In 2010 Wireless InnovationConference and Product Exposition (SDR’10)., Washington D.C., USA, Dec 2010.

1

Nucleus Project

Flexible & Energy Efficient Wireless Communication Transceiver DesignFlexible & Energy Efficient Wireless Communication Transceiver Design

Cognitive & Software Defined RadiosCognitive & Software Defined RadiosCognitive & Software Defined RadiosCognitive & Software Defined Radios Design Issues and Project GoalsDesign Issues and Project GoalsDesign Issues and Project GoalsDesign Issues and Project Goals

� Goals:� Identify novel methodology and develop tools for design of future SDRs & CRs� Find efficient and flexible architecture implementations� Investigate algorithms that enable efficient architectures

RF

Fro

nte

nd

(Do

wn

-/Up

con

vers

ion) Tx Modulator

Inner Receiver

Tx Framing & FEC

Outer Receiver

Layer 1 SW

Cell Search

DelayProfile

Estimation

Physical Layer Scheduling & Ctrl Layer 1 SW

Data L1 Config/Ctrl System Information/Higher Layer Ctrl

Layer 2/3 Stack (MAC, RLC, RRC)

AFC

DL Power Control

Source: Infineon

RF

Fro

nte

nd

(Do

wn

-/Up

con

vers

ion) Tx Modulator

Inner Receiver

Tx Framing & FEC

Outer Receiver

Layer 1 SW

Cell Search

DelayProfile

Estimation

Physical Layer Scheduling & Ctrl Layer 1 SW

Data L1 Config/Ctrl System Information/Higher Layer Ctrl

Layer 2/3 Stack (MAC, RLC, RRC)

AFC

DL Power Control

Source: Infineon

Opt

.Li

mit

Implementation Efficiency

Por

tabi

lity

(1/P

ortin

g E

ffort

) Optimization

adding Flexibility

105..106

Opt

.Li

mit

Improvement of Design Methodology

Tough Real-Time Constraints(Throughput, Latency)

Portability/Flexibility vs. Efficiency

Cognitive Wireless Networks and Radios:

� Must sense or be cognitive of the environment� Other user interference, multi-path, noise, etc.� Time-variations

� Must be intelligent to analyze the situation andfind the optimal communications protocol, frequency band, transmission mode, etc.

� Must reconfigure

� And constantly adapt to changing mobile environments

Find the best protocol, frequency band,and transmission mode.

Software Defined Radios:

� SDRs provide the required flexibility, hence are the main enabler of cognitive radios and networks

� Energy efficiency and challenging performance requirements demand Heterogeneous Multi-Processor System-on-Chip architectures.

AlgorithmsAlgorithms

Performance vs. ComplexityPerformance vs. ComplexityPerformance vs. ComplexityPerformance vs. Complexity

ArchitectureArchitecture

Future Transceiver DesignFuture Transceiver DesignFuture Transceiver DesignFuture Transceiver Design

and transmission mode.

IdentifyIdentify

Nuclei:Nuclei:Nuclei:Nuclei:critical, demanding, algorithmic kernel that captures common functionalities within and/or among transceivers

SWSWSWSW

ProcessorProcessorProcessorProcessor(GPP, ASIP,…)(GPP, ASIP,…)(GPP, ASIP,…)(GPP, ASIP,…)

ConfigurationConfigurationConfigurationConfiguration

ASICASICASICASIC…

Algorithms:Algorithms:Algorithms:Algorithms:� sphere decoder� channel decoder

� channel estimator� MAC processing� ….

PE2(rASIP)

PE3(DSP)

Comm. Arch.

PE1(ASIP)

Implementations provided by Flavors Implementations provided by Flavors bundled with Processing Elementbundled with Processing Element

Flavor

PE

ToolsTools

Novel Design MethodologyNovel Design MethodologyNovel Design MethodologyNovel Design Methodology

BoardSupportPackage

Nuclei

Transceiver Description

Mapping &Evaluation

PE2PE1 PE3 PE4 PE5

N 1 N 2 N 7 NNN 5

FlFl

Fl

Fl NNI

Compile

N 1

N 7 N 5N 2Non N Tasks

NucleusLibrary

NIFl Flavor & PE

PE2(rASIP)

PE3(DSP)

MEM

Comm. Arch.

PE5(FPGA)

PE1(ASIP)

PE4(GPP)

HW Platform

Performance vs. ComplexityPerformance vs. ComplexityPerformance vs. ComplexityPerformance vs. Complexity

Physical-Layer :

� Most standards specify a minimiumrequired error rate performance

� Achieving an error rate better than the specified is not necessary.

Investigate trade-offs between algorithm complexity and

energy efficiency for achieving the necessary

performance constraints.

Standard required error rate

DVB-T 2x 10-4 coded BER

802.16 10-6 coded BER

802.11n 10% PER

Example MIMO Demapper:

Ultra High-Speed Mobile Information and Communication

Heterogeneous MPSoC architecture made of efficient, but flexible components (Flavors bundled with PEs).

MEM

Comm. Arch.PE5(ASIC)

PE4(GPP)

User inputs

N1 N2 NN1

N3

Waveform description

.cpn

Config & constraints

.xml .xml

Flow inputs

Nucleus Library

FlavorLibrary

Board SupportPackage

.xml

Platform Description

VPU-basedVirtual Platform

Multi-threaded hostimplementation

Code Generator

Nucleus Mapper

ISS-basedVirtual Platform

� Tool Flow Implementation

� Advanced Simulation Techniques

This includes investigation of critical functions and their implementations, e.g.

� X-hundred Mbit/s MIMO-OFDM Demappertape out together with ETH Zürich

ZF : zero forcingMMSE : minimum mean-squared errorOSIC : ordered successive interference cancellation SD : sphere decoderES : exhaustive search

SD

OSIC

Complexity

Err

or

Rat

e

ES

ZFMMSE

Algo. X

Efficient algorithms: Efficient algorithms: Min. complexity Min. complexity

(error rate)(error rate)for certain error rate for certain error rate

(complexity)(complexity)

Example MIMO Demapper:

CrossDisciplinaryResearch

� 6 Professors� around 20 Researchers

Contact:Torsten.Kempf@iss.rwth-aachen.deJeronimo.Castrillon@iss.rwth-aachen.de