Spectrum Holes Sensing Policy For Cognitive Radio Network Using Reinforcement-Learning
Divya Kumari · 2014
Cognitive radio is a promising technology that allows Secondary (unlicensed) Users (SU) to access and share the frequency band originally allocated to Primary (licensed) Users (PU). Secondary nodes are cognitive radio and primary users have license to use spectrum. The secondary nodes utilize the spectrum whenever it is free or not occupied by primary user. Whether the spectrum is occupied or not is found by sensing techniques, like energy detection or cyclostationary etc. Energy consumption for spectrum sensing depends on techniques used in secondary nodes. If Secondary node follows periodic or random policy, energy consumption would be high or spectrum holes missing high respectively. And so the secondary user needs optimized spectrum sensing policy to detect spectrum holes. The main problem is, if the player does not have prior knowledge about the reward distributions of the different machines, it is obviously impossible to derive optimal action selection policies. The proposed sensing policy users use the Reinforcement Learning Algorithm (RL) in artificial intelligent techniques to learn behavior of primary users to predict when the spectrum is occupied and not. It provides energy sensing policy and also low probability of miss spectrum holes detection. The proposed system to improves the throughput of the secondary user and improves the energy efficiency while controlling the miss detection probability.