Research On Link Initialization And Spectrum Allocation Technology For Cognitive Networks

In the past two decades, the communication technology has advanced rapidly, and more and more new communication products have appeared in people’s daily lives. In particular, the popularization of wireless network has made people’s study, work and entertainment more convenient. In themeantime, people also has increasingly higher requirement for the wireless spectrum resources, and in some open unlicensedband, the spectrum resource is always "in short supply". However, the current spectrum allocation policy as adopted fixed spectrum assignment principle, and no matter how low the utilization ratio of users on licensed spectrum is, the unlicensed users cannot use these spectrum resources. The fixed spectrum allocation principle can help manage the spectrum resources, but it has also caused the current "deficiency" of spectrum resources. Without reassigning the spectrum resources, in order to conflict between the "deficiency" of spectrum resources and the low utilization ratio of fixed spectrum, we need a new spectrum sharing technology to balance the use of spectrum resources. Guided by this demand, the cognitive radio technologyemerged at the right moment. The popularization of cognitive radio equipment can help improve the utilization ratio of spectrum resources and solve the conflict of uneven use between licensed and unlicensed spectrum. Different from the traditional wireless communication network, the cognitive network is a dynamic and heterogeneous communication network, and due to these characteristics, the practical application of cognitive network will face various challenges. By studying various aspects such as cooperative spectrum sensing, secondary user communication link initialization and spectrum allocation in the cognitive network, a series of solutions for the cognitive network are proposed. The main works are outlined as follows:In order to obtain accurate spectrum sensing results, we study the cooperative spectrum sensing, and propose a cooperative spectrum sensingalgorithm based on reliable decision. First of all, in accordance with the characteristic that thcreliability of local decision result will increase with the received SNR (signal to noise ratio), we calculate the boundary of using single decision threshold for decision. When the SNR is high, the single decision threshold is used for decision; when the SNR is low, the double threshold is used for decision, which can increase the decision accuracy. If thestatistics of the energy accumulation lies in the credible region, deliver the decision result to the fusion center; if thestatistics of the energy accumulation is at thedoubt region, the decision result obtained from the optimal single threshold is unreliable. Use the log likelihood ratio of the received signal to verify the unrealizabledecision result, and the verified decision result will be sent to thefusion center. In order to ensure the performance of system detection, we propose a data method that uses the previous experience as a combined weight.In order to solve the link initialization problem of secondary users in asymmetric model of cognitive networks, we study the channel rendezvous method between transmitter and receiver, and propose the rendezvous algorithm based on the asymmetric model of cognitive networks. First of all, we analyze the asynchronous time slot communication system. Then, according to the characteristic of rendezvous sequences with variant prime length, we propose a method which can dividedifferent types of finite prime sets with co-prime basic sequence lengths. This method is used to define the basic sequence length of transmitter and receiver, which can ensure that 2 sequences will make guaranteed rendezvous. In order to reduce the redundancy of channel hopping sequences, a method is proposed to use the available channel set to directly generate the channel-hopping sequence. Through this method, the basic length of formed rendezvous sequence will adapt to the available channel number. When the secondary users have very few available channels, it can effectively reduce the rendezvous time. In order to improve the utilization ratio of rendezvous sequence, we propose the strategy to use the "blank" slots in rendezvous sequence. When the "blank" slots are used for channel rendezvous, it can form a subsequence that can improve the rendezvous efficiency.In order solve the link initialization problem of secondary users in symmetric model of cognitive networks, we study the universal channel-hopping sequence generation method with rendezvous ability, and propose the rendezvous algorithm based on symmetric model of cognitive networks. During the research, first of all, we analyze the rendezvous between quick-slowsequences and quick sequences with different jump steps, and then prove the necessity for them to have rendezvous. When the secondary user has equivalent available channel sets, by designing the ID sequence that satisfies the asynchronous rendezvous conditions, ditferent types of basic sequences can correspond to each other at the same lime slot. The ID sequence is used to connect three basic sequences, and form the hopping sequence unit of rendezvoussequence. When the secondary user has nonequivalent available channel sets, by changing the basic sequence generation parameters, it can realize centralized rotation of rendezvous channel in available channel, in this way to ensure that the secondary user will make guaranteed rendezvous within a cycle. In order to improve the utilization ratio of rendezvous sequences, we propose the strategy to use the "blank" time slot in available channel, and the strategy proposed in this paper can effectively improve the rendezvous opportunities between secondary users. In order to evaluate the rendezvousperformance of proposed algorithm, we conduct simulation of rendezvous scenario of secondary users in symmetric model of computer, and compare the proposed algorithm with other similar algorithms. The simulation results show that proposed algorithm outperforms other algorithms in rendezvousperformance, and the short-hopping sequence unit used in this paper can effectively shorten the time-to-rendezvous.In order to solve the link initialization problem of secondary user quipped with multiple cognitive device, we study the construction method for parallel channel-hopping sequences with rendezvous ability, and propose the rendezvous algorithm facing multiple cognitive devices.First of all, we analyze 3 kinds of rendezvous models used in multiple cognitive devices. Then, we prove the rendezvous performance between quick-slow basic sequences with different lengths. Based on that, we put forward the method touse the available channel set to directly generate the parallel channel-hopping sequence. By analyzing the relationship between the rendezvous time and other essential factors, including the basic sequence length and the quick-slow sequence allocation parameters, we propose an optimized basic sequence allocation strategy. This strategy can be used to further reduce the rendezvous time between multiple cognitive devices. In order to obtain the state information of license channel in a timely manner, we add "bland" time slots to the parallel rendezvous sequence, so that the secondary users can conduct cyclic spectrum sensing.According to the activity ofprimary users, the secondary users can adjust the length of spectrum sensing sequence by themselves, in this way to ensure the rendezvous efficiency of parallel rendezvous sequence.In order to solve the centralized spectrum allocation problem, we propose the parallel spectrum allocation algorithm based on artificial physics optimization. Firstly, we analyze the centralized spectrum allocation scenario, and the interference graph model is established by defining the interference distance. In order to connect spectrum allocation with interference constraints, the interference graph is transformed into matrix distribution model. With the goal to maximize the network throughput, we try to find an optimal allocation plan for allocation vectors. Because it is difficult to obtain the optimal solution, we use the artificial physics optimization method to search for the suboptimal solution. By decomposing the search space, the allocation vector is transformed into allocation sub-vectors, and we conduct parallel search of all sub-vectors at the same time.After the search process ends, we choose the optimal allocation results to generate the optimal solution. The population initialization method of evolutionary algorithm is improved to increase the search efficiency. The fitness of the initial population can be improved by reducing random operation in feasible solutions, which can ensure that the subsequent search can start at the better "position". In order to improve the fairness of allocation, we have improved the operation to remove the interference constraint, so that all allocation dimensions will have an equal opportunity in participating in the channel allocation. To prevent early particle aggregation during iterative search, which might cause the suboptimal solutions fall into local optimal solution, the diversity control mechanism is introduced into the artificial physics optimization method. During the particle movement, the distance between particles in the population is measured, and the directional function is used to control the "contraction" and "expansion" of particle swarm.