Decentralised detection using censored observations is an effective way to mitigate communication constraints where each sensor only transmits the informative' observations to a fusion centre (FC), and censors tho...
详细信息
Decentralised detection using censored observations is an effective way to mitigate communication constraints where each sensor only transmits the informative' observations to a fusion centre (FC), and censors those deemed uninformative'. In this study, a censoring rule is used at each spatial diversity channel such that only the observation exceeding a local threshold is transmitted, where the local threshold controls the communication rate. In particular, the authors consider a scenario where there exists a communication constraint at the FC. To design a proper communication bandwidth allocation (CBA) algorithm, two Ali-Silvey distance measures, i.e. the J-divergence (JD) and Bhattacharyya distance (BD), are utilised as objective functions to achieve a manageable optimisation procedure for designing local thresholds. The CBA algorithm is proposed by solving a constrained non-linear programmingproblem through the sequential quadratic programming method. Besides, the detection performance is analysed under the censoring scheme, given the optimised local thresholds. Numerical results indicate that the two Ali-Silvey distance-based algorithms can outperform the uniform bandwidth allocation strategy;meanwhile, the BD-based method performs better at high signal-to-noise scenarios but slightly worse at low, compared with the JD-based method.
暂无评论