鐘陽晶++梁茹冰++黃小虎
摘 要: 為了降低基于接收信號強度指示(RSSI)測距誤差對節(jié)點定位的影響,解決RSSI測距定位誤差較大的問題,提出基于RSSI高斯濾波的最小二乘支持向量回歸機LSSVR定位算法(LSSVR?GF?RSSI)。LSSVR?GF?RSSI算法先利用高斯函數(shù)濾除誤差較大的RSSI值,篩選出較準確的RSSI值,再依據(jù)這些值計算未知節(jié)點離錨節(jié)點間的距離。將這些距離作為LSSVR的輸入,建立基于RSSI測距的LSSVR定位算法模型,最終,估計未知節(jié)點的位置。仿真結(jié)果表明,提出的LSSVR?GF?RSSI算法能夠有效地降低均方定位誤差,比傳統(tǒng)的基于RSSI的LSSVR定位算法減少了約12%~20%。
關(guān)鍵詞: 接收信號強度; 最小二乘支持向量回歸機; 高斯函數(shù); 定位; 無線傳感網(wǎng)絡(luò)
中圖分類號: TN914?34 文獻標識碼: A 文章編號: 1004?373X(2017)11?0006?04
LSSVR wireless sensor network location algorithm based on Gaussian filter RSSI
ZHONG Yangjing, LIANG Rubing, HUANG Xiaohu
(College of Mathematics and Informatics, South China Agricultural University, Guangzhou 510642, China)
Abstract: In order to minimize the influence of range?finding error of received signal strength index (RSSI) on node localization, and solve the problem of big location error existing in localization algorithm based on RSSI range?finding, a least?squares support vector regression location algorithm based on Gaussian filter RSSI (LSSVR?GF?RSSI) is proposed. The LSSVR?GF?RSSI algorithm uses the Gaussian function to filter the RSSI values with big error, and screen out the accurate RSSI values. According to the above values, the distance between the unknown node and anchor node is calculated. The distance is used as the input of LSSVR to establish the LSSVR location algorithm model based on RSSI range?finding to estimate the location of unknown node. The simulation results show that the LSSVR?GF?RSSI algorithm can reduce the mean square localization error effectively, which is 12%~20% lower than that of the traditional LSSVR localization algorithm based on RSSI.
Keywords: received signal strength; least?square support vector regression; Gaussian function; localization; wireless sensor network
0 引 言
無線傳感網(wǎng)絡(luò)(Wireless Sensor Networks,WSNs)系統(tǒng)[1?2]主要應(yīng)用于人為力量無法到達的復(fù)雜區(qū)域事件的監(jiān)測和數(shù)據(jù)的采集與傳輸[3]。而采集的數(shù)據(jù)的實用性與其地理位置息息相關(guān)。獲取沒有準確位置的信息是毫無價值的。然而,在WSNs網(wǎng)絡(luò)中,多數(shù)傳感節(jié)點隨機部署,并且多數(shù)節(jié)點位置是未知的[4]。由于只有已知空間位置的感應(yīng)數(shù)據(jù)才有實用價值,故須利用定位技術(shù)估計傳感節(jié)點的位置。
受硬件條件和無線環(huán)境因素的制約,在WSNs中對傳感節(jié)點的定位仍是一項挑戰(zhàn)工作。目前,已提出多類定位算法[5?6]。依據(jù)定位過程是否需要測距,可將這些算法劃分為測距定位、非測距定位。前者表示在估計未知節(jié)點位置時需要直接估算未知節(jié)點離錨節(jié)點間的距離,即測距;而后者是通過利用整個網(wǎng)絡(luò)的連通性估計未知節(jié)點的位置。因此,通常測距定位算法精度優(yōu)于非測距定位算法。
常用于測距定位算法中的測距策略有:信號到達角度AOA(Angle of Arrival)、到達時間TOA(Time of arrival)、基于接收信號強度RSSI(Received Signal Strength Index)。其中基于RSSI測距是利用未知節(jié)點接收到來自錨節(jié)點發(fā)射信號的強度估算路徑傳播損耗,進而估計未知節(jié)點離錨節(jié)點間的距離。由于基于RSSI測距無需額外的硬件設(shè)備,其廣泛應(yīng)用于低成本的無線傳感網(wǎng)絡(luò)WSNs中[7?8]。因此,研究并尋求高精度的RSSI測距算法具有重要的實用價值。
文獻[9]提出基于RSSI值校驗的未知節(jié)點定位算法。依據(jù)錨節(jié)點對未知節(jié)點影響力的不同,設(shè)置不同的加權(quán)因子,同時擇優(yōu)選擇優(yōu)質(zhì)的錨節(jié)點參與未知節(jié)點的位置估計。文獻[10]提出基于RSSI校正的WSNs定位算法。先利用高斯函數(shù)篩選較準確的RSSI值,再對這些RSSI值設(shè)定加權(quán)系數(shù),進而估計未知節(jié)點的位置。文獻[11]提出基于LSSVR的無線傳感網(wǎng)絡(luò)定位算法。引用最小二乘支持向量回歸機LSSVR提高定位精度。支持向量回歸機SVR(Support Vector Regression)依據(jù)統(tǒng)計學習理論,在非線性回歸估計中具有優(yōu)良的性能,即使在小樣本環(huán)境,也表現(xiàn)出較好的泛化能力[12]。
為此,結(jié)合高斯函數(shù)的篩選特性以及LSSVR在統(tǒng)計學習方面的優(yōu)勢,提出基于RSSI高斯濾波的LSSVR無線傳感網(wǎng)絡(luò)定位算法(Least?Squares Support Vector Regression location algorithm based on Gaussian filter RSSI,LSSVR?GF?RSSI)。LSSVR?GF?RSSI算法先利用高斯函數(shù)選擇偏差較小的RSSI值,再將這些RSSI值參與測距,將這些測距向量作為LSSVR的輸入,進而估計未知節(jié)點的位置。仿真結(jié)果表明,提出的LSSVR?GF?RSSI算法能夠有效地降低均方定位誤差。
4 結(jié) 論
本文針對基于RSSI測距定位精度低的問題,分析測距原理以及影響定位誤差的因素,并提出基于RSSI高斯濾波的最小二乘支持向量回歸機LSSVR定位算法LSSVR?GF?RSSI。LSSVR?GF?RSSI算法利用高斯函數(shù)濾除偏差較大的RSSI值,即選擇較準確的RSSI值,利用這些值轉(zhuǎn)化為距離,然后將這些距離作為LSSVR模型的輸入,最終估計未知節(jié)點的位置。仿真結(jié)果表明,與LSSVR?RSSI算法相比,提出的LSSVR?GF?RSSI算法有效地降低了均方定位誤差,且沒有增加額外的運行時間。
參考文獻
[1] BULUSU N, HEIDEMANN J, ESTRIN D. GPS?less low cost outdoor localization for very small devices [J]. IEEE personal communications magazine, 2012, 7(5): 28?34.
[2] LANGENDOEN K, REIJERS N. Distributed localization in wireless sensor networks: a quantitative comparison [J]. Computer networks, 2013, 43(4): 499?518.
[3] 王越,周奧,劉金城.無線傳感器網(wǎng)絡(luò)中非測距混合定位算法[J].傳感器與微系統(tǒng),2015,34(2):147?151.
[4] 江禹生,馮硯毫,管芳,等.無線傳感網(wǎng)非測距三維節(jié)點定位算法[J].西安電子科技大學學報(自然科學版),2012,39(5):140?148.
[5] NICULESCU D, NATH B. DV based positioning in Ad Hoc networks [J]. Telecommunication systems, 2003, 22(1): 267?280.
[6] SAVARESE C, RABAEY J M, LANGENDOEN K. Robust positioning algorithms for distributed Ad?Hoc wireless sensor networks [C]// Proceedings of 2002 the General Track of the Annual Conference on USENIX Annual Technical. Berkeley: USENIX, 2002: 317?327.
[7] PATWARI N, HERO A O, PERKINS M, et al. Relative location estimation in wireless sensor networks [J]. IEEE transactions on signal processing, 2003, 51(8): 2137?2148.
[8] OUYANG R W, WONG K S, LEA C T. Received signal strength?based wireless localization via semidefinite programming: noncooperative and cooperative schemes [J]. IEEE transactions on vehicular technology, 2010, 59(3): 1307?1318.
[9] QING X, GOH C K, CHEN Z N. Impedance characterization of RFID tag antennas and application in co?design [J]. IEEE transactions on microwave theory technology, 2009, 57(5): 1268?1274.
[10] 文春武,宋杰,姚家振.基于RSSI校正的無線傳感器網(wǎng)絡(luò)定位算法[J].傳感器與微系統(tǒng),2014,33(12):134?138.
[11] 張曉蓮,唐加山.基于改進RSSI測距的LSSVR三維WSN定位算法[J].電視技術(shù),2014,38(19):131?135.
[12] STUTZMAN W L. Estimating directivity and gain of antennas [J]. IEEE antennas propagation magazine, 1998, 40(4): 7?11.
[13] 何艷麗.無線傳感器網(wǎng)絡(luò)質(zhì)心定位算法研究[J].計算機仿真,2011,28(5):163?166.
[14] SODERSTROM T, STOICA P. System identification [M]. London: Prentice?Hall, 1999.
[15] 趙吉文,劉永斌,蘇亞輝.新型直線電機支持向量機非線性建模研究[J].光學精密工程,2006,14(3):450?455.