Introduction

The Multiple-Carrier Multiple-Antenna Systems (MCMA) group emphasizes on wireless communication system design and wireless communication algorithm development with practical considerations. The current focus is the design and implementation of a multiple-carrier multiple-antenna transceiver system targeting 1Gbps data rate in a standard WLAN 20-MHz channel. In order to provide 1Gbps throughput, we are faced with the challenge of providing robust high data rate link in the limited radio spectrum. This calls for the development of sophisticated communications and signal processing techniques that provide spectrally efficient communication links. One approach to achieve these goals is to use multiple-antenna communications.

Fig. 1 illustrates the three major benefits of multiple-antenna transmission in a typical indoor scenario. By using combined diversity and spatial multiplexing the data rate and robustness against channel fluctuations can be increased as well as interferers suppressed. Theoretical and simulation studies have characterized multiple-antenna algorithms under various channel models but less attention has been paid to implementation issues including architectural and circuit optimization, feasibility for integration on silicon, and testing in real indoor wireless environment.

The goal of this research is to bring system design together with architecture and algorithm design and technology to obtain a better understanding of the key trade-offs in the implementation of wireless communication systems. Furthermore, the MCMA group aims to develop a multiple-antenna testbed that will allow wireless experiments and characterization of various multiple-antenna algorithms under realistic channel conditions.

Fig 1: Combined diversity and spatial multiplexing.

System Overview

Using multiple carriers like in Orthogonal Frequency Division Modulation (OFDM) is an efficient transmission scheme for fading channel with the elimination of complex channel equalizers as required by single-carrier systems. OFDM is a modulation choice for many WLAN standards. In order to increase the data rate of conventional OFDM systems, multiple-antenna processing can be used on each narrow-band sub-carrier so that the capacity increase is proportional to the number of antennas. Communication theory claims that multiple-antenna transmission provides diversity and spatial multiplexing gains simultaneously, but there is a trade-off between how much of each type of gains a system can achieve. Conceptually, multiple-antenna systems split the encoded data into parallel streams that are simultaneously transmitted in the same frequency band.  As a result, the receiver has a difficult task to simultaneously distinguish all multiple data channels thus requiring large computational complexity. A number of multiple-antenna processing techniques are considered including transmit and receive beamforming, MMSE, and SVD. Developed by BWRC researchers, adaptive SVD is an optimal multiple-antenna algorithm for slow-varying fading channel through the use of a feedback link between the transmitter and receiver. Fig. 2 shows the block diagram of the envisioned multiple-carrier multiple-antenna system with the digital baseband processing running on the BWRC prototyping platform BEE connected to a parallel RF front-end.

Fig 2: Block diagram of the multiple-carrier multiple-antenna system.

Prototype Platform

Implementing the algorithm that provides best theoretical channel capacity might not always be feasible due to enormous computational complexity under real-time operation constraints or due to large energy consumption. Our approach for implementation is to explore architectures of various multiple-antenna algorithms, identify common building blocks, and choose solutions that are the most suitable for evaluation and test using our fast prototyping test platform.

The test platform used is the Berkeley Emulation Engine (BEE) which is based on 20 Xilinx FPGAs capable of supporting up to 600 billion operations per second. In addition, the same top-level algorithm description is used to retarget the design for an ASIC implementation, so very little additional effort is necessary to map the design into a silicon chip that is guaranteed to work. System blocks are designed and implemented in Simulink. This leads to highly modular and visualized design and allows fast Simulink-to-hardware mapping using the BWRC BEE and SSHAFT tools.

Fig 3: 1-Mbps narrow-band transmission system running on the BEE prototype platform.

Radio Front-End

A multiple-antenna RF front-end will be connected to BEE, where all baseband algorithms are implemented in order to establish wireless transmissions. We are designing a parallel RF front-end that is scalable up to 16 antennas.

Many design parameters need to be considered for a multiple-antenna front-end: good matching of all components, common synthesizer and clocks, etc. To get a good understanding about the influence of all these parameters in multiple-antenna systems and for algorithm development, the analog impairments are first modeled on BEE. Our approach combines analysis of analog impairments in the design of RF front-end through modeling, simulation and reference design measurements of a single antenna front-end. Finally, a parallel RF front-end will enable us testing of multiple-antenna algorithms and synchronization schemes under various indoor channels and realistic analog hardware impairments. 

Fig 4: Exemplary 2.4-GHz front-end used for our wireless experiments.