Research On QoS Guarantee Over Distributed Cognitive Radio Networks

The inefficient spectrum utilization led by t he administrative management makes the cognitive radio(CR) become a key technology in the next generation of wireless networks. Based on the sensing results, secondary user(SU) accomplishes the Qo S requirements, meanwhile satifying the primary user ’s(PU’s) collision limitations. However, there are still several chanllenges in the practical distibuted cognitive radio networks(DCRNs) to optimize the SU’s Qo S performance. First, compared to the PU with high priority on the spectrum usage, SU only is provided with low priority on spectrum utilization. The unequal priority of channel usage between SU and PU promotes SU’s spectrum handoff action. Second, SU with the security problem is the same as any communication system. Third, there is no availabe spectrum resource over DCRNs. This paper focuses on the mentioned three challenges.(1) To guarantee the Qo S performance of two types of video in the spectrum handoff process over DCRNs, an evolution handoff strategy combining the mixed integar non-linear programming(MINLP) and partially observable Markov decision process(POMDP) is designed. Considering the limitation factors(such as imperfect PU’s sensing results, limited observation capability, the non-synchronization between SU and PU, SU’s mobility and SU’s hardware limitation) in DCRNs, the optimization problem in spectrum handoff process is fomulated as the combination of short-term optimization and long-term optimization. When the real-time video(or video stream) is transmitted, the short-term optimization is formuated as MINLP, based on the the specific limitations factors(e.g. PU’s allowable collision probablity, SU’s mobility and so on). Furthermore, considering the imperfect sensing results and limited sensing capability, SU formulates the long-term spectrum handoff as the POMDP model to design the evolution handoff strategy, meanwhile optimizing the long-term Qo S performances of real-time video and video stream.(2) Considering the greedy cognitive radio(GCR) revises the MAC protocol parameters to adopt the small backoff window(SBW) attack, the proactive protection scheme is proposed. In the proposed scheme, GCRs with high priority are allowed to access spectrum resources, while satisfying two prior rules: Rule 1, GCRs are required to prolong the transmission time of each packet in exchange for the high priority on spectrum usage; Rule 2, GCRs allow normal SUs(NSUs) to transmit data with themselves simultaneously in certain slots. Meanwhile, the proactive protection scheme for NSUs consists of two steps, which are recording GCRs’ transmission time with a counter and maximizing their throughput through the nonlinear programming with constrained condition. Furthermore, considering the joint MAC(JMAC) attack consists of objective function attack and SBW attack, the process of JMAC attack and anti-JMAC countermeasure is modeled as the composite Markov decision process(CMDP). Based on this CMDP model, a novel reinforcement learning proactive(RLP) protection scheme is proposed for NSUs to efficiently avoid GSUs’ JMAC attack. Meanwhile, the NSUs’ average throughput in each slot is also maximized.(3) Considering that there are available channels or not over DCRNs, two antenna selection strategies are proposed to guarantee SU’s Qo S performance based on the TCP Reno protocol. Assuming that there are available channels, the TCP throughput maximization is formulated into an max-min optimization problem according to the TCP Reno protocol. Furthermore, based on the solution of the proposed problem, the antenna selection strategy for maximized TCP throughput(AS-MTT) is proposed. Moreover, since there is few available channels over DCRNs, based on the opportunistic spatial orthogonalization scheme, the antenna selection strategy for maximized practical TCP throughput(AS-MPTT) is designed, while considering PU’s interference tolerance. Finally, Using the AS-MTT and AS-MPTT, SU is able to adopt various modulations and antenna selection technique to adjust the transport-layer Qo S. Meanwile, the available time-frequency-space resource is efficiently utilized.