摘要
提出了多包接收时的信号场强叠加模型,建立了观测场强与传感器位置的映射关系。由于传感器数量相对于网格数量是稀疏的,将传感器定位转化为压缩感知问题求解,以减少观测的信号数量,并提出了NL1-norm算法计算出传感器的位置。通过数值仿真,分析了传感器信号功率、观测信号数量以及传感器个数对定位误差的影响。相同条件下,验证了NL1-norm算法的定位精度相比最小化L1-norm算法和贪婪匹配追踪(GMP)算法提高了2倍。低信噪比情况下比较得出,基于CS的节点定位方法误差和观测代价都明显小于RSSI和MDS-MAP方法。
This paper proposes the superposition model of signal strength when MPR (multi-packet reception) is used in WSN, and es- tablishes the mapping relation between the observed signal strength and sensor location. Because of the sparsity of the number of the sensors relative to the number of the grids, the issue of sensor localization can be transformed into a compressive sensing issue, therefore the number of the measured signals is reduced. The NL1-norm algorithm is proposed to calculate the sensor position. With numerical simulation, the influence of the sensor signal power, the number of the measured signals and the number of the sensors on localization error is analyzed. It is verified that the localization accuracy of the NL1-norm algorithm is two times better than those of the minimized Ll-norm and greedy match pursuit (GMP) algorithms. Comparison under the low SNR circumstance shows that the localization method based on CS has less location error and measurement cost than RSSI and MDS-MAP methods.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2014年第6期1201-1208,共8页
Chinese Journal of Scientific Instrument
基金
国家自然科学基金(61370088)
国家国际科技合作专项项目(20140FB10060)资助