Multilayer LMS interference suppression algorithms for CDMA wireless networks
Savo Glisic, Zorica Bečanović Nikolić, Bojan R. Dimitrijevic, Graeme Woodward · IEEE Transactions on Communications · 2000
The general theory of adaptive self reconfigurable interference suppression schemes is applied to several specific practical problems mainly suppression of m-level m amplitude-shift keying, m phase-shift keying (PSK), and m quadrature amplitude modulation signals. This is a practical situation when a code-division multiple-access (CDMA) network is overlaid with standard microwave systems. Another example is a multirate CDMA network where a limited number of high bit rate CDMA signals are allowed to use much higher power level due to lower processing gain. The algorithm is well suited for a modular software radio concept, which we believe, will be more and more accepted in future wireless communications. Further modifications of the schemes necessary for these applications are described, and numerous results are presented to illustrate performance improvements. A general interpretation of these techniques based on so-called multilayer least mean squares (LMS) algorithm is introduced and discussed. The algorithm is based on estimating fast changing interfering signal parameters by using parallel structures, which are fast but complex. At the same time, estimation of slow-varying signal parameters over a large range is accomplished by using an LMS algorithm that is simple but slower. In this way, suppression of the interference occupying the same bandwidth as the CDMA signal is possible with reasonable implementation complexity. For this case, a BER0 dB for binary PSK interference, J/S>17 dB for 8-PSK, J/S>27 dB for 32-PSK, and J/S>40 dB for 128-PSK.