Application of Artificial Intelligence in Cognitive Radio for Planning Distribution of Frequency Channels

The technology of cognitive radio is an innovative technology designed to increase the efficiency of radio frequency channels. The most important part of this technology is the allocation of frequency channels, the search for free and busy channels (sensing) in the process of transmitting traffic. We studied the search for free and busy channels based on channel properties to form clusters. The initial data of the model is a two-dimensional distribution of frequency channels, which is characterized by the fact that each user from a certain finite set has a fixed number of frequency channels with the same bandwidth. The clustering of free and busy channels is investigated in the assumption of random distribution of free and occupied channels for each user. It has been shown that numbers of free and used channels approximately coincide and their sum is 1/3 of the total number of channels, so the number of uncertain channels was 2/3. It is also shown that applying of machine learning for frequency resources based on this model is able to reduce the time of channel search by 10%. The generated statistical model of clustering channels is applied to the algorithm of machine learning, namely state vector machine (SVM). This algorithm is applied to channels that transmit VoIP traffic. The Markov 2-character sequence of free and occupied states of one channel with a statistical connection between two consecutive time intervals is applied for modelling of traffic. It is shown that the model provides the correct definition of clustered channels at an acceptable level (not less than 50 %) for up to 10 iterations of the algorithm without the construction of a new vector state machine. For VoIP traffic, this means that the SVM vector must be rebuilt once in 0.2 seconds at a bit rate of 12.2 Kbps.