Considering the initialization of the CMAC weight memory cells the selection of the training samples, and a quantization method of the input variable, the faster CMAC learning algorithms, variable resolution of non-overlapping receptive field is proposed , which can be conveniently realized in hardware Programming Logic Curcuit.

The critical overlap concentration (C) and the critical entanglement concentration C_B of linear polystyrene (LPS) and branched polystyeene (BPS) have been disseussedThe values of C and C_E were calculated according to the Graessley and Klein equation C=0.

The networks are divided into a number of overlapping clusters and produce gateway nodes,using connected dominating set theory to optimize with the topology structure of networks,selecting the gateway nodes with higher energy to be new dominating nodes,the network load and energy consumption are balanced,coming to the purpose of extending the life time of networks.