Robust Codes for Space-Time Communication, Iterative Decoding, and Multi-Rate Broadcast

University Of California-Los Angeles, Los Angeles CA

Investigators

Abstract

The primary goal of this research is to produce practical codes that provide unequaled robustness, approaching the performance predicted by compound channel information theory for linear Gaussian channels. The resulting channel codes will find application in many scenarios including wired and wireless broadcast, wireless local area networks, asymmetric digital subscriber lines, military anti-jam communications, and frequency hopped communications. Results from compound channel information theory state that a single code exists that supports communication over every channel that has a mutual information above the information rate of the code. Thus, once the transmitter power spectrum is fixed, a code need not be specialized to a particular frequency selective channel in a single antenna system or a particular space-time path-gain matrix in a multiple antenna system. While this theoretical result is more that thirty years old, practical code design has focused to date on a weaker notion of robustness, that of average performance according to a statistical channel model, such as Rayleigh fading. The power of the code design approach taken in this research is that the code design is independent of statistical assumptions about the channel. The code simply works on every channel that it possibly could. A second goal of this research is to develop codes that support multiple rates to different users. Here again, information theory (this time information theory for degraded broadcast channels) lays out the limits of performance. Our research will closely approach these performance limits by applying turbo codes to the concept of superposition coding for the degraded broadcast channel.