Generalized-Bi-Connectivity For Fault Tolerant Cognitive Radio Networks

Cognitive radio gets more and more attention, which is called "The Next Big Thing" inwireless networking. Fault-tolerant topology is a basic requirement in wireless network. Incognitive radio networks (CRNs), available channels of cognitive users dynamically changesince a channel becomes unavailable when the channel is reclaimed by primary users.Therefore, fault tolerance of CRNs highly depends on change of channel availability.However, traditional definition of bi-connectivity concerns only node failure and is notsuitable to CRNs. In this study, we introduce a new definition of generalized-bi-connectivity(g-bi-connectivity). A CRN is said to be g-bi-connected if the remaining network stillconnected when any one of the two events occurs: i) any node fails; ii) any channel becomesunavailable. Based on this definition, we solve problems as below:i) Spectrum prediction scheme for cognitive radio: we propose a SW-HMM (SlidingWindow-Hidden Markov Model) based on traditional HMM (Hidden Markov Model), we usegeneric to train the parameters of this model. The simulation results shows that, ourSW-HMM is better than ANN and Markov model, which are used in related works.ii) Build g-bi-connectivity topology network: we give the definition of g-bi-connectivity,and build a g-bi-connected network by assigning power and channels to the cognitive users.Our objective is to minimize the maximum transmission power of users and the number ofchannels required. We propose a two-stage approach which consists of power assignmentstage and channel assignment stage. We prove that the maximum transmission power ofcognitive users is optimized and derive an upper-bound on the number of channels required inthe approach.iii) Topology recovery: we propose two algorithms to keep g-bi-connectivity of thenetwork while node-leaving or node-joining.iv) Simulation: we do much simulation for the algorithms mentioned above, to prove theeffect of these algorithms.