基于压缩感知的被动式移动目标轨迹测绘  

Compressive Sensing Based Device-Free Moving Target Trajectory Depiction

作　　者：王举[1] 陈晓江[1] 常俪琼 房鼎益[1] 邢天璋[1] 聂卫科 

机构地区：[1]西北大学信息科学与技术学院,西安710127

出　　处：《计算机学报》2015年第12期2361-2374,共14页Chinese Journal of Computers

基　　金：国家科技支撑计划项目基金(2013BAK01B02);国家自然科学基金(61170218;61272461;61373177);西北大学研究生创新教育项目(YZZ13104;YZZ14002)资助

摘　　要：被动式目标轨迹测绘以其无需目标携带任何设备的优点吸引着许多应用,如野生动物监测、入侵安全监测等.针对现有基于被动式目标轨迹测绘方法,因频繁定位而导致计算开销大和大量观测数据导致通信能耗高的问题,文中提出基于压缩感知的被动式目标轨迹测绘(Compressive Sensing Based Device-Free Target Trajectory Depiction,CSTD)算法,仅用少量观测数据一次性精确测绘出目标轨迹,减少了计算和通信开销,降低了能耗.文中的关键性发现及CSTD优点是:(1)轨迹上不同的位置及其估算具有时间独立性和空间统一性,可将不同位置映射到统一的物理空间一次性测绘出目标轨迹,避免传统方法频繁定位计算开销的问题;(2)目标轨迹与监测区域的空间位置相比具有稀疏性,利用压缩感知原理通过少量观测数据就能精确测绘出目标轨迹,降低了数据量和能耗.为适应实际应用中的大规模场景需求,该文给出了可扩展的CSTD算法模型,并提出了目标轨迹稀疏度未知(目标经过的位置数未知)下的稀疏恢复算法.部署了48个节点的8m×8m真实实验,结果表明在降低观测数据量的同时,CSTD较现有经典算法至少提高了63%的轨迹测绘精度.Without relying on devices carried by the target,device-free trajectory depiction is attractive to many applications,such as wildlife monitoring and asset protection in industrial facilities.To deal with the challenges,such as repeatedly positioning calculation,high data volume and the energy consumption,existed in most of current device-free trajectory depiction methods,this paper introduces a novel compressive sensing based device free target trajectory depiction（CSTD）method.The key observation is that：（1）the location estimation has the time-independence and spatial-uniformity.Different locations can be mapped into a unified physical space and depict target trajectory one time,thus,avoid the repeatedly positioning calculation;（2）the target trajectory has the sparse property,which can apply the compressive sensing to track the target accurately even with just a few measurements,thus,we can reduce the data volume and the energy consumption.More importantly,CSTD can be easily scaled to work in large areas.We also design a sparse recovery algorithm which can recovery the signal without requiring the sparse level.Results from the real-world deployment of 48 nodes in an 8m×8marea demonstrate that CSTD can provide an improvement of 63% accuracy.

关 键 词：被动式跟踪(定位) 压缩感知 数据量 物联网 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]