Optimal utilization of frequency spectrum in wireless networks particularly in device to device communication is of significant importance owing to the growing demand. Traditional methods to optimal spectrum utilization of spectrum are not sufficiently efficient and result in loss of spectrum. Recently, application of Cognetive radio is suggested to solve this problem. Cognetive radio is a smart wireless system which is aware of the spectral traffic condition of its environment in an instantaneous way and through these spectral conditions, changes the power of transmitter and the type of modulation and it adapts to the environment. The main purpose of this paper is to investigate the problem of spectral sharing. Today, communication systems suffer from main problems including limited bandwidth, download speed increase, rate increase and saving in transmitted power. To solve such problems, new methods based on machine learning in spectrum sharing are necessary to overcome such challenges. In this work, using cellular learner automata, a method is proposed for simultaneous assigning of spectrum and resource. The aim of each pair of transmission is to transmit in an appropriate channel and power level so that it can maximize its compensation in cellular learner automata. In these scenarios, compensation is taken as the difference between operational (collective) and consumed power. The cost of the consumed power is the signal to interference noise ration. Proposed method is simulated on a LTE-A network as well as an NS2. Proposed algorithm is of rapid convergence and semi-optimal efficiency in low repetitions. تفاصيل المقالة