Interference suppression is one of the important functions for mobile communications and software radio. First, this paper shows a new type of interference suppression method by P-RCE (Probability-Restricted Coulomb Energy) which is applicable to mobile communications and software radio. P-RCE is one of the neural networks and mainly used in the field of pattern classification. Secondly, this paper presents several characteristics of this method. For example, it is found from our studies that good suppression effects can be performed even when the interference signals exist closely adjacent to the desired signal and/or total number of signals is more than that of the antenna elements. Next, this paper discusses two types of improvement of processing speed for new suppression method. One is the setting up the learning and non-learning intervals, and the other is the restriction of the number of prototype cells. According to the results, fairly good improvement is realized.

For usage in fading environments, the idea of utilizing radio signal processing with software programmable devices promises greater flexibility and functionality. However, to realize these benefits very high performance DPSs and AD converters are required, especially for adaptive antennas with CDMA. One potential method of reducing the implementation complexity is to employ multicarrier (MC) modulation as it allows parallel signal processing in addition to its benefits in combating fading and ISI effects. This paper proposes a system utilizing MC DS/CDMA and a recursive adaptive array antenna processing algorithm for estimating the array response vector (array manifold) and hence the optimal weights. In order to evaluate this system performance results obtained through computer simulations are presented. The proposed MC system improves the accuracy of the direction-of-arrival (DOA) estimation of the signal by 2 degrees as compared with a single carrier system. It is also shown that the proposed adaptive array antenna algorithm for MC DS/CDMA reduces the number of iterations of the power method significantly and allows parallel processing of the adaptive algorithm.