Cooperative detection over multiple parallel channels: A principle inspired by nature

We consider an information theoretic model that describes parallel reception of a common signal by many receivers. The signal is distorted by noise and independently decoded by each receiver. Receivers cooperate in the sense that the outcome is conveyed to a decision center, which makes the final decision on the received signal by a majority vote. A potential scenario which could be described by this model is a cluster of base stations jointly receiving signals from a mobile station and conjoining individual decodings. Similar principles apply for cognitive radio setups where a secondary user employs many weak subcarriers, subject to sudden drop out by occupation from the primary users. The whole system is modeled as a cascade of channels, and quality of information exchange is measured by mutual information. This allows for modeling and optimizing quantization and detection in a unifying approach. Numerical evaluations demonstrate that the increase of mutual information by using additional channels is only logarithmic. 4-QAM is investigated as a concrete example and system performance is numerically investigated. The model in this paper is motivated by one used for describing information exchange in biological neural networks, revealing so called stochastic resonance by adding noise to signals.