Ant intelligence inspired blind data detection for ultra-wideband radar sensors

Given the computational complexity and sophisticated implementation of traditionally parametric channel estimators, it has been gradually recognized that the existing data detection methodologies based on the finite impulse response (FIR) propagation channel modeling may become infeasible for ultra-wideband (UWB) radar sensors, especially in some large-scale distributed scenarios. By exploiting the implicit information involved in the received signals, in this investigation, we present a non-parametric UWB data detection scheme for the distributed radar sensor networks. A novel characteristic representation is suggested first. From a pattern classification point of view, a group of quantitative features are then extracted by making full use of the inherent property of UWB propagations. Thus, UWB data detection is formulated as a pattern classification problem in a multidimensional feature space. By thoroughly utilizing the self-similarity of the representative patterns, the ant swarm intelligence inspired clustering algorithm, with the new designed ant movement strategy, is adopted to perform unsupervised data detections. The developed scheme is independent of any a priori modeling information, which essentially avoids the expensive parametric estimators and thus enables practically feasible realizations. To alleviate the computational burden, the principle component analysis (PCA) is further employed to compress the feature space. The simulation results validate the new algorithm, which is superior to the other popular non-parametric data analysis schemes.