Enhanced Spectrum Planning in Cognitive System Based on Reinforcement Learning

článek ve sborníku ve WoS nebo Scopus

angličtina

This paper presents preliminary results of interference less spectrum planning which is performed by reinforcement learning. The frequency planning of the wireless services is very difficult in current overfilled spectrum situation since it is nearly impossible to find spectral hole worldwide to deploy a new wireless service. The possible solution is an open dynamic spectrum access which could be implemented as a part of cognitive radio. Moreover, the modern wireless standards such as LTE-A partly implement cognitive radio improvements, e.g. carrier aggregation system, which enables using unused parts of the frequency spectrum to decrease interference and increase data throughput. It could be realised both the intra-band and the inter-band solution. According to the measured data of the spectrum situation in various environments, we prepared best case of channel switching in LTE-A and WI-FI systems, which is based on the reinforcement learning to minimize interference with primary users represented by measured data. Using this technique, we are capable to obtain very low misdetection probability and large variety in channel switching.

cognitive radio, spectrum sharing, spectrum survey

URBAN, R.; HUTOVÁ, E.; NEŠPOR, D.

2013

15. 8. 2013

978-1-934142-26-4

Proceedings of PIERS 2013 in Stockholm

759

762

4