BLIND ADAPTIVE ALGORITHMS FOR OPTIMAL MINIMUM VARIANCE CDMA RECEIVERS

The present invention utilizes a filter that works on an algorithm that filters out multiuser and multi-path interference. The algorithm is programmed into the DSP processors of receiver handsets and base stations in cellular and multihop wireless radio networks which use the CDMA technology. It can also be implemented in hardware. Additionally, the invention works in a blind manner such that it learns without requiring training samples. The algorithm can be expressed as updating the receiver parameter vector f(n) as:f(n+1)=[I−C(CHC)−1CH][f(n)−mfy(n)yH(n)f(n)]+C(CHC)−1g(n)wherein y(n) is a data vector for each transmitted bit W1(n) from user 1, the receiver will estimate that bit by filtering y(n), i.e.w1(n)=fH(n)y(h),where H denotes Hermitian; g(n) corresponds to the multipath parameters and is updated by a separate revision; and C is a matrix constructed from elements of the desired user's spreading code. Importantly, the filter based on this algorithm is adaptive and can be used blind, i.e. no training of the device is necessary prior to operation, rather, the device is improved each time the algorithm is run.