协作通信网络中基于协作概率的负载均衡机制

在协作通信存在的网络中,针对已有的负载均衡技术没有考虑用户切换后协作概率的不足,在已有的负载均衡技术基础上,利用目标小区负载状况得到切换后的协作概率,与小区信号强度和负载状况相结合,设计出应用于协作通信网络中具有负载均衡功能的切换判决函数,提高切换判决的精度。仿真表明该机制在均衡网络负载的同时,提高了系统容量和公平性。

基于概率协作表示的几何保持图嵌入

为了充分获取高维数据的几何结构特征、增强高维数据在子空间中的识别能力,基于概率协作表示拥有潜在判别信息和数据几何信息的能力,提出一种基于概率协作表示的几何保持图嵌入(PCRGPGE)的降维新方法.在新方法中,相似高维数据样本的概率协作表示相似,且其概率协作表示的重构在低维子空间中也保持相似,从而使得高维数据所蕴含的低维结构信息保持在子空间中.在PCRGPGR中,首先对高维数据进行概率协作表示并重构,以获取数据的潜在判别和几何信息;然后,采用高维数据的类内和类间的构图方式,得到数据的几何分布信息和判别信息.将PCRGPGR与PCA、LDA、LPP等10个图嵌入方法在2个经典的人脸图像数据集PIE29和IMM上进行对比试验.结果表明:所提出的PCRGPGE是一种有效的降维方法,能在低维空间较好地保持高维数据的内在结构信息和增强模式判别能力.

基于概率协作表示的运动想象脑电分类算法

在脑机接口的研究中,针对运动想象脑电信号的特征识别,提出了一种基于概率协作表示的分类方法(probabilistic collaborative representation based classification,ProCRC),通过比较测试样本在每个类别中的最大可能性,从而确定其所属的类别。采用BCI竞赛数据库Ⅲ中的数据集Ⅰ,利用S变换进行特征提取,然后对不同的分类器进行比较,以分类准确率作为评价标准验证了该算法的有效性。该算法的分类准确率能够达到90%,为脑机接口系统分类算法的研究提供了新思路。

基于单演特征和概率性协作表示的人脸识别

针对光照、遮挡、姿态、样本数有限等问题对人脸识别的挑战,提出一种融合单演特征和概率性协作表示的人脸识别方法 .概率性协作表示分类方法(ProCRC)相对于稀疏表示分类方法(SRC)和协作表示分类方法(CRC)具有明显的分类优势.但在样本数有限情况下ProCRC方法中样本需更精确地被表示才能获得精确的结果.针对该问题,提出使用单演特征来精确地表示人脸样本.实验结果明表所提出方法能够有效地提高在复杂光照、遮挡、姿态、样本数有限条件下的人脸识别率.

基于概率协作表示的多表情序列融合识别

传统表情识别往往是基于单一图像进行特征提取、训练及识别,缺乏在动态时间上的细微表情变化提取。利用时间前后的人脸表情变化信息,提出了一种基于概率协作表示的多视频序列融合表情识别方法。先采用主动外观模型(AAM)定位出人脸表情的68个特征点,利用提出的融合策略将相邻3帧表情图像的AAM特征进行融合,最后利用概率协作表示的分类优势进行识别。实验结果表明,在CK+表情数据库上,所提出的方法能够把握表情的时间变化信息,相比于近几年的表情识别算法,在识别率上取得了较好的效果。

Outage performance analysis of multiple users in cognitive networks

A cooperative model of multiple primary and secondary users coexisting cognitive network is presented. In this model, the control center is aware of all the users＇ locations in order to allocate the nearest secondary user to the primary user. The control center is aware of the information of the unused spectral resources in terms of the feedback of the sensing results from the secondary users. It allocates idle frequency bands among the secondary users. The primary user accesses the base station （BS） in orthogonal subchannels, and it cooperatively transmits packets with the secondary user and exploits the free band assigned by the control center to amplify-and-forward what it receives immediately. Under this scenario, the outage probability of the cooperative transmission pair of the primary and secondary transmitters is derived. The numerical simulation of the outage probabilities as a function of primary transmission probability ps, power allocation ratio ξ between the primary and secondary users, and the numbers of the primary and secondary users are given respectively. The results show that the optimal system performance is achieved under the conditions of ξ=0.5 and the numbers of the primary and the secondary users being equal.

Relay selection region and system performance analysis for interfered cooperative ad-hoc networks

The performance of interfered cooperative ad-hoc networks is analyzed by stochastic geometry analysis and a selection region of relay is presented. First, assuming that the distribution of nodes in the random network follows the Poisson point process （PPP）, a closed-form expression of the outage probability is derived for the best relay selection （BRS） scheme. Secondly, the capacity of the network is presented for this scheme. Finally, a performance factor is defined to evaluate the performance gain obtained from the BRS. By using this factor, a relay selection region is found to guarantee the performance gain from the BRS. The analysis and simulation results show that the performance of the BRS not only depends on the densities of source nodes and relay nodes but also on the factors of networks such as the path loss factor and the decoding threshold. And the BRS has a greater advantage than direct transmission （DT） in hush environments such as the long transmission distances, much interference and the high decoding thresholds.