A Software Simulation Testbed for CDMA Wireless Systems

A Software Simulation Testbed for CDMA Wireless Systems Vishwas Sundaramurthy Electrical and Computer Engineering Department and April 26 th 1999

Wireless cellular systems u Tremendous growth in coverage – – Mobility – u More economical than wired-line systems -- infrastructure Capacity increases Code Division Multiple Access(CDMA) – Efficient usage of channel capacity – Robust against defects in the transmission and channel

Challenges in designing CDMA systems u Communications research – performance – usage of capacity improvements in: u Advanced signal processing techniques – channel estimation – multiuser detection – coding u A large number of design choices – complex environment (channel models) – different algorithms for blocks in the system

Motivation for a simulation testbed u What are the channel properties or “environmental” factors to deal with? • noise • multi-path delays • attenuation • fading

Motivation for a testbed (contd. ) u What is the best combination of algorithms ? u How do these algorithms behave with one another? u Need for performance data before investing in infrastructure

The concept of a CDMA Testbed u u Algorithms – evaluation – tradeoffs System parameters – channel models – user dynamics u Performance indicators – Bit error rate (BER) – Frame error rate (FER)

The concept of a CDMA Testbed (contd. ) Mobile phone + Base station u u Algorithms – Uplink – Downlink System parameters Prototyping

Contributions u Flexible wireless CDMA multiuser link – Developed in the Simulink and Matlab Environment – Algorithms n channel estimation n multiuser – detection System configuration n channel models n different spreading methods

The CDMA concept 1 Spreading waveforms chip 1 -1 1 1 -1 -1 + Spread bit sequences + Superimpose 2 +3 +1 3 -1 bit period -3 Baseband waveform

CDMA works!! Spreading u Nc=7 (spreading matrix) -1 -1 -1 -1 1 1 1 1 -1 -1 1 x . . 1. . -1 -1 -1 1 1 -1 (code matched filtering) 1 1 1 x -1. . 3. . -1. . (superimposed chips) Detection 1 1 -1 1. . -1. . 2 consecutive bits of 3 users u . . -3. . 1 -3 3 1 -3 1 1 1 -1 1 3 -1 -1 (baseband signal) 11 5 11 -3 -9 -7 (soft decision) 1 1 -1 -1 (hard decision)

CDMA Wireless Link User’s data bits SOURCE CODING CHANNEL CODING multiple users SPREADING TRANSMITTER Detected bits of all the users MODULATION noise DECODER DETECTOR CHANNEL ESTIMATOR RECEIVER DEMODULATION

W-CDMA Simulation Testbed in Software u u Matlab environment Simulink block diagrams

CDMA Wireless Link User’s data bits SOURCE CODING CHANNEL CODING multiple users SPREADING TRANSMITTER Detected bits of all the users MODULATION noise DECODER DETECTOR CHANNEL ESTIMATOR RECEIVER DEMODULATION In Simulink

CDMA Wireless System Testbed Simulink Version Chip matched filter Multiuser Detection AWGN Channel Error Rate Calculation User_Data Wireless Channel Chip MF Multiuser Detector Decorrelating Detector Error Counter Channel Estimation User Data Max. Likelihood Channel Est. Show Stats Update Parameters Statistics

Features u Flexible wireless CDMA multiuser link – better modeling of a real system u Clean user interface – vary parameters of the system u Modular – library of different algorithms u State-of-the-art algorithms – multiuser detection – channel estimation

Development Methodology C - Code MATLAB Code Simulink Workstation Simulink Library Algorithms Wrapper Display and analysis of data

System Development using Simulink C - MEX S - function SIMULINK Library MATLAB S - function Block BLOCK A BLOCK B data in BLOCK C data out BLOCK E u u BLOCK D Simulation framework: discrete time-steps Data-flow based simulation

Algorithms u Spreading methods – Gold Codes – M-Sequences – Random spreading codes u Channel estimation – maximum likelihood Multiuser detection – matched filter – decorrelator – differencing multistage u

Maximum Likelihood Algorithm Received chip matched filter output ri = ( U Z) bi + ni Shifted spreading matrix and channel impulse response Noise Preamble bits Estimate by Maximum likelihood method

Multiuser detection u Code matched filter Cross correlation matrix u Decorrelating detector u Multistage detector (lth stage) T yi = ( U Z ) ri T R = ( U Z ) (U Z) -1 ddecorr = R yi <l> < l-1 > zmultistage = yi - R A d < l-1 > d = sign (zmultistage)

Modularity CDMA library Channels Detectors Associated components Multiuser Matched-Filter Detector x Multipath Channel Matched-Filter Detector b cdma-tx-gen u Library of different algorithms u Plug and play! x Multiuser Decorrelating Detector Multipath Wireless Channel b Decorrelating Detector cdmatx Rd wksp Single user receiver chan i/p code MF o/p sync delay chip MF o/p Multiuser Multistage Detector preamb_wksp Multistage Detector bitconvert

Graphical User Interface u Control parameters – SNR – spreading gain – # users – bit length u Change modules / algorithms – spreading – multiuser detection

Automated “Net-List” Generation u The Simulink model is reconfigured based on parameters – add/remove user data blocks depending on {number of users} – change algorithms

The segment under consideration Mobile Transmitters K Asynchronous Users data bits Spreading and modulation Spreading • spreading gain Detected bits of K users detection Demodulation and chip matched filtering • spreading type AWGN noise synchronization Multi-User (Base-station) Receiver Channel parameters ( SNR, #paths )

Spreading, Channel and Chip Matched Filtering spreading channel spreading x spreading Chip waveform . . data bits chip matched filter vector

Spreading, Channel and Chip Matched Filtering 1 Spread bit sequences 1 -1 Superimpose Baseband waveform 1 -1 -1 Channel output waveform +3 +1 -1 -3 Add Noise Channel output waveform Chip matched filter output

Traditional method of simulation modeling u Represents baseband signals explicitly – superimposed chips – AWGN noise – baseband signal u Problem – baseband signals are continuous time signals – a simulation system models them as a discrete time signal – finite number of samples leads to an error – slow simulation speed

Traditional Method Spreading, Channel modeling and Chip-matched filtering bits-1 sprd Spreading Function-1 Sample and Hold Tx-delay-1 AWGN T z-1 AWGN channel bits-2 sprd Spreading Function-2 Tx-delay-2 Pulse Generator S/H Discrete-Time Integrator Chip Matched FIlter

The efficient method of simulation modeling u Does not represent intermediate signals explicitly u All the “blocks” in the baseband segment collapsed into one entity – spreading, channel model and matched filtering – input n bit streams n spreading, channel parameters – provides discretized channel waveform (matched filter output)

Efficient Method Spreading, Channel modeling and Chip-matched filtering bits-1 Unit Delay 1 z sprd_cmf bits-2 spreading channel and chip matched filter

Simulation time speed-up

Analysis of simulation time from profiling

Simulink Test-bed Demo

Simulation in Software and DSP hardware Simulink Block Diagrams Real-Time Workshop (RTW) TI C 67 EVM (Evaluation Module)

Prototyping Methodology C - Code Matlab Code Block Diagram Libraries Algorithms Display and Analysis of Data Simulink Workstation RTW generated C - Code With RTW support for DSP hardware DSP Code Generation Tools DSP hardware

RTW support for C 67 EVM 3 x 3 1 Simulink Model Matrix Constant U V 3 x 3 sdspmmult U U*V Matrix Multiplication mult. mat 2 To File 1 1 Matrix Multiply V Matrix Constant 1 Real Time Workshop • Create “template make file” • Change references to cross-compiler (TI C 6 x code generation tools) • Generate code C 67 DSP hardware • Download to DSP board • Analyze data in Simulink / Workstation

Future Work - CDMA Library of blocks u Fading channel and supporting estimation and detection algorithms u Channel parameter estimation and tracking – Maximum Likelihood – Subspace u Source and Channel coding u Power control

Future Work - DSP support u Execution of software test-bed on DSP hardware – u Real-Time Workshop (RTW) support for C 67 boards Use multiprocessing on DSPs to execute the Simulink based system

Summary u u u A Multi-user Wireless CDMA Link – backbone to grow a bigger system Flexible system – vary parameters – state-of-the-art algorithms from the library Simulink/Matlab Environment – GUI for changing parameters and algorithms – data analysis and profiling – RTW for DSP code generation