ABSTRACT
This chapter introduces an adaptive averaging cell (AVG) structure tailored for realistic environments with nonhomogeneous variability. It demonstrates the efficiency of using adjustable weights in the averaging cell structure taking into account the heterogeneous variability levels among the replicas. Associated with fault-tolerant techniques based on redundancy, the AVG graphically calculate the most probable value of a binary variable from a set of error-prone physical replicas. The chapter presents the AVG architecture and its main features, and obtain the equation to calculate the output error probability. It introduces the idea of unbalanced voting with the aided design–AVG technique and demonstrate that it is possible to optimize the values of the average weights in order to minimize the output error probability. The chapter also present the results of Monte Carlo simulations of the AVG structure with balanced and optimal unbalanced weights and compare the behavior of both approaches in the presence of the heterogeneous variability levels in the input replicas.
