Research Project Title
Low Complexity, High Data Rate Wireless Architecture

Principal Investigators
Sean Meyn
Ada Poon

Unit # 32
Project Overview

This project will bring together the disparate and highly complementary expertise of the two PIs to address topics in receiver design for mobile or fixed wireless broadband channels. The starting point is based on recent results in channel coding: For any of the channel models considered in academic or industrial research, the random code optimizing an error exponent is obtained using a distribution with finite support. Moreover, under very general conditions the same conclusion is obtained for the distribution achieving capacity. What do these results really mean in practice? The application of a simpler signal constellation design has a range of benefits beyond quick detection. Consider the following issues in communication design:

 

Dynamic range  Large signal bandwidth has obvious benefits, but also several drawbacks: As signal bandwidth grows, so does inter-symbol interference as well as interference from other users. Moreover, an increase in bandwidth requires more stringent dynamic-range requirements on the analog-to-digital converter (ADC). For example, the ADC in an 802.11g WLAN SoC consumes 198 mW. The entire receiver consumes 315 mW, so that more than half of the power goes to the ADC. Power dissipation in the ADC scales approximately linearly with the number of levels in the signal constellation. That is, the ADC power will be reduced by half when the number of levels is reduced by half.

 

Decoding complexity  Higher data rates and better range requirements motivate the use of MIMO antenna systems. There are three major types of space-time decoding algorithms: linear, such as Alamouti schemes, iterative such as MMSE decision feedback, and maximum-likelihood ML) such as spherical decoding. The ML decoding algorithm drastically outperforms the other two approaches. However, the complexity of ML algorithms increases as a polynomial function of the number of signal constellation points.

 

Optimal signal constellations are typically sparse. Consequently, on choosing an optimal con-

stellation we obtain a reduction in decoding complexity.

 

We will to develop signal constellation design methods to simultaneously improve the reliability of receivers, reduce dynamic-range requirements and hence power consumption, and reduce the demodulation/decoding complexity of the receiver for mobile or fixed wireless broadband access. We will also extend recent results in [6, 7] to MIMO antenna channels, and develop practical algorithms to construct signal constellations as well as efficient ML-based space-time decoding schemes.