Research On Spatiotemporal Spectrum Sensing And Frame Optimization Technology In Cognitive Radio

With the tremendous growth of the number of wireless communication devices and services,the requirement of spectrum resource is increasing sharply,leading to the shortage of spectrum resources.But existing experiment researches have shown that the allocated spectrum is inefficiently utilized.That's because in the fixed spectrum allocation policy,licensed users have exclusive right to use the allocated spectrum.But the utilization of spectrum is dynamic.As one of the effective solution to these problems,cognitive radio technology allows the unlicensed users to access the temporary unoccupied spectrum dynamically on the premise of not making any effect on the communication performance of the licensed users.By this way,the spectrum efficiency can be increased.Spectrum sensing is the key task of cognitive radio.Only when the available spectrum opportunity is detected as existence or nonexistence in certain band,can the unlicensed users make corresponding adjustment to the following communication processes.This thesis focuses on improving the detection accuracy and efficiency of spectrum sensing.The design of spectrum sensing algorithm,the selection of cooperative nodes,data fusion,performance evaluation and the parameter optimization problems are studied.The main work and contributions of this thesis are summarized as follows:1.Since cognitive users located in different locations may experience different spectrum reuse opportunities,a spatiotemporal sensing scheme based on location estimation is proposed.First,the location and transmission power of the licensed user's transmitter are estimated via spatial sensing.The information can be used to estimate the cognitive users' maximum interference constrained transmission powers at different locations,and to select a few nodes that are closer to the licensed user's transmitter for temporal sensing.The observations obtained by the selected user in a period of time are simply merged to reduce the reporting overhead.Considering the correlation of the observations of different cognitive users,the local statistics are weighted by spatial correlation of selected nodes rather than all nodes.The simulation results show that the proposed scheme improves the probability of detecting spectrum holes and achievable capacity compared with traditional pure spatial or temporal sensing scheme.2.Two cooperative spectrum sensing schemes based on log likelihood ratio(LLR)are proposed by taking account of correlated cooperation users with probable unequal signal-to-noise ratio(SNR)values in practical scenario,where the reporting channel is fading.At first,the location estimation of licensed user's transmitter is obtained via spatial sensing,and some users are selected as temporal sensing users rather than all based on three kinds of user selection rules,that is,users that are closer to licensed user's transmitter,users with minimum correlation,and users randomly selected.In order to alleviate the problem of frequent data exchange and reduce the amount of reporting data in traditional soft combing,two local test statistics derivation schemes are proposed,and the resulted average LLR of samples or the LLR of average sample is taken as local detection outcome.The decision user adopts equal gain combing or local statistics-weighted fusion to combine selected users' local detection outcome and then makes the final decision.Through the comparison and analysis of the detection performance of the schemes with different user selection criterions and fusion rules in different correlation scenarios,the optimum scheme can be selected.Then the parameters of the number of sensing nodes,samples and antennas are optimized jointly to achieve the balance between the detection accuracy and overhead.3.A new adaptive weighted spectrum sensing scheme based on full duplex for dynamic traffic of licensed user environment is proposed.The frame structure of simultaneous sensing and transmission is adopted to avoid the tradeoff between sensing accuracy and throughput and the additional burden on synchronization between licensed user and cognitive user brought by conventional frame structure,which inserts a short quiet period into the beginning of each frame.In practical scenario,licensed user may arrive and departure randomly.A weighted sensing scheme using a power function based on the corresponding sampling sequence is designed,where heavier weights are assigned to the latest samples to reveal the actual licensed user's state in near real time.The power exponent is further adjusted to the sensing situations to achieve lowest false alarm probability under a certain detection probability constraint.Then,an analytical model considering all possible state transitions of licensed user is developed to evaluate achievable interference,outage,throughput and energy efficiency.Furthermore,the optimal frame duration yielding both optimal false alarm probability and throughput is computed.After that,a fast search algorithm is proposed to track the opti-mal duration.The results indicate that the proposed scheme can achieve lower false alarm probability and higher energy efficiency over a wide SNR range than that of the existing weighting schemes without increasing the order of the computational complexity.