Research On The Key Technologies Of Spectrum Management In Cognitive Radio Networks

In order to improve the utilization of wireless spectrum resource further, secondary users(SUs) have been allowed to access spectrum holes of primary users(PUs) opportunistically in cognitive radio networks(CRNs). Before using spectrum holes, SUs obtain the right of using them by doing a deal with PUs. Due to the lacks of both private information and its prior information(So called incomplete information) between each other in practical spectrum trading processes, decision-makers(including PUs and SUs) need to be able to make decisions with incomplete information. From this point of view, this dissertation does research on the problem of how to make decisions rationally with incomplete information in market models, and the detail contents mainly include the following aspects:1. When a PU uses the second-price sealed bid auction model to sell the idle slots of its spectrum among SUs, the SU who sends the highest bid to the PU will be the winner in the auction and needs to pay the secondary highest bid to the PU for the right of using the idle slots. However, in some situations, the differences between the highest bids and the secondary highest bids are larger, which will have a negative effect on the PU’s profits obtaining from the auction process. In order to increase the PU’s profits, the PU can set reserve prices during the auction process. A decision-making model which just utilizes the historical information of the highest bids to set the optimal reserve prices has been proposed in this dissertation to solve the problem of how to set reserve prices with incomplete information. Based on the results of Monte Carlo simulation experiments,the proposed decision-making model can not only help the PU raise its profits, but also almost have no negative effect on the utilization of spectrum holes.2. When multiple PUs sell the idle slots of their spectrum by auction mechanisms at the same time, each SU who is a rational decision-maker must consider the problem of how to select the auctioneer with the maximal winning probability(the optimal auctioneer). A synchronous selection strategy which is based on the multi-auction bandit game model and a sequential selection strategy which is based on optimal stopping theory have been proposed in this dissertation to solve the problem of how to select auctioneers with incomplete information. In the multi-auction bandit game model, Exp3 online learning algorithm has been used to help each SU find its optimal auctioneer. Due to Exp3’s larger maximal averaged regret during the auctioneer selection process, this dissertation has proposed a new online learning algorithm which just utilizes the historical information of each PU’s won bids, and has proved the maximal averaged regret of the new algorithm is less than that of Exp3 in theory. The sequential auctioneer selection problem has been transformed to a classical secretary selection problem in this dissertation. The intelligent SU who is a SU with ability to make decisions sequentially estimates the winning probability in each auction and needs to make a decision between to stop at the current auction and to continue to estimate the winning probability in the next auction. Based on the results of Monte Carlo simulation experiments, each SU can find its optimal auctioneer with incomplete information by both of online algorithms in the multi-auction bandit game model. At the same time, the maximal averaged regrets of both of online algorithms are finite and the maximal averaged regret of the secondary online algorithm is obviously less than that of the first online algorithm. In the sequential selection model, the auctioneer selection strategies which are based on the optimal stopping rules are effective and can satisfy the intelligent SU’s different requirements.3. There must be a situation in which one PU carries out a spectrum transaction with one SU in practical CRNs. Bargaining theory which belongs to the domain of game theory is a powerful tool to study this situation. In the bargaining process, the problem of how to send a rational price to the opponent under the condition that both PUs and SUs have no knowledge of each others’ both private information and its prior information must be investigated. This dissertation has proposed a decision-making model in the bargaining process without any both private information and its prior information of each other in opportunistic cooperation networks. Based on the results of Monte Carlo simulation experiments, the proposed decision-making model can not only let the bargainers(both the PU and the SU) reach an agreement on the prices in the bargaining process, but also bring two bargainers into a win-to-win situation, which means it is effective and practical.