Loading...

Cross layer Design in Cognitive Radio Networks

Majd, Arash | 2013

471 Viewed
  1. Type of Document: Ph.D. Dissertation
  2. Language: Farsi
  3. Document No: 45262 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Pakravan, Mohammad Reza
  7. Abstract:
  8. In this thesis, the challenges of the PHY and MAC layers in cognitive radio networks are discussed and a plan which considers the cross layer concept between them is presented.In the first part of the thesis, a brief survey on various spectrum sensing techniques is presented. A detector which can improve the network performance is introduced and based on this detector, the important problem of trade-off between spectrum sensing period and the network throughput is considered. For this purpose, first of all, the truncated SPRT detector is presented and for low SNRs the optimal truncation time which leads to maximum network throughput is obtained. In the second part of the thesis,the opportunistic spectrum access strategies are discussed. Two important parameters in obtaining the optimal strategies are the network occupation statistics and the other SUs presence. Channel occupation scheme of PU network, is generally assumed to be Markovian or Bernoullian. First, the networks with Markovian channel occupation scheme are considered and using the complex framework of Dec-POMDP, the decision trees for each SU is obtained and using the dynamic memory bounded technique, the omplexity of this technique is reduced considerably. Meanwhile an algorithm which can produce decision trees appropriate with QoS equired in each SU is also presented. In the other part of the thesis, an SU network with Bernoulli model, is considered and a multi-user channel acces strategy in the presence of channel impairments like noise and fading is proposed. For this purpose, the problem is considered in the format of an optimization problem which tries to obtain SUs access strategies so that the total throughput is maximized and on the other hand, the probability of collision with PU network in each channel is guaranteed to be less than a predefined value of ζ. At the end, a heterogeneous Bernoullian network is considered and it is proved that the optimal strategy in this network is to assign a channel to each SU with probability one. Obtaining an optimal strategy in these networks needs complete channel state information which is not practical. Hence, a distributed algorithm based on graph coloring technique is proposed and through simulation, it is shown that the proposed algorithm outperforms the similar algorithms
  9. Keywords:
  10. Cognitive Radio ; Markov Decision Making ; Spectrum Sensing ; Sequential Detector

 Digital Object List

 Bookmark

No TOC