Abstract：In view of gas source localization in wireless sensor networks, an adaptive projected subgradient method（APSM）is firstly introduced to estimate the gas source position. However, the APSM for locating is difficult to converge and consumes much time because of the concentrations corrupted by noise. So a distributed adaptive deflection projected subgradient method（DADPSM）is proposed for modifying the iteration search direction and combing the characteristic of distributed calculation in WSN. Under the attenuation model of a gas source in the wind field, this algorithm could process the distributed information by DADPSM, the deflection subgradient direction is used in place of original gradient and the deflection subgradient projection hyperplanes is applied as the searching areas in the process of relaxed projection to achieve the gas source localization. Simulation proves that this algorithm can provide good convergence property, locate the gas source accurately, and save large amount of energy.
［1］匡兴红,邵惠鹤.基于WSN的两种气体源定位算法研究［J］.仪器仪表学报.2007, 28(2):298-302.
［2］Wang H, Zhou Y M, Yang X L, et al. Plume source localizing in different distributions and noise types based on WSN［C］//2010 WRI International Conference on Communications and Mobile Computing, Shenzhen, China, 2010:63-66.
［3］朱勇,王萍,张昕明.基于WSN的加权组合三边气体源定位算法研究［J］.电子测量技术.2009,32(8):12-15.
［4］Wang C L, Wu D S.Decentralized positioning and tracking based on a weighted incremental subgradient algorithm for wireless sensor networks［C］//The 68th IEEE Vehicular Technology Conference, Calgary, AB, Canada, 2008:1-5.
［5］Sundhar R S, Nedic A, Veeravalli V V. Distributed subgradient projection algorithm for convex optimization［C］//IEEE International Conference on Acoustics, Speech and Signal Processing Proceedings, Taipei, Taiwan, 2009:3653-3656.
［6］Blatt D, Hero III A O.Energy-based sensor network source localization via projection onto convex sets［J］.IEEE Transaction on Signal Processing, 2006, 54(9):3614-3619.
［7］Yukawa M, De Lamare R C, Yamada I.Robust reduced-rank adaptive algorithm based on parallel subgradient projection and krylov subspace［J］.IEEE Transactions on Signal Processing, 2009, 57(12): 4660-4674.
［8］Yukawa M, Slavakis K, Yamada I.Multi-domain adaptive learning based on feasibility splitting and adaptive projected subgradient method［J］.IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences, 2010, E93-A（2): 456-466.
［9］Cavalcante R L G, Yamda I, Mulgrew B.An adaptive projected subgradient approach to learning in diffusion networks［J］.IEEE Transactions on Signal Processing, 2009, 7(57): 2762-2774.
［10］Fukazawa Y, Ishida H.Estimating gas-source location in outdoor environment using mobile robot equipped with gas sensors and anemometer［C］//IEEE Sensors 2009 Conference-SENSORS, Christchurch, New Zealand, 2009:1721-1724.
［11］Crank J.The mathematics of diffusion［M］.Oxford, UK: Oxford University Press, 1956:11-271.
［12］Zhang Y, Wang L.A particle filtering method for odor-source localization in wireless sensor network with mobile robot［C］//Proceedings of the 8th World Congress on Intelligent Control and Automation, Jinan,China,2010:7032-7036.
［13］Ampeliotis D, Berberidis K. Low complexity multiple acoustic source localization in sensor networks based on energy measurements［J］.Signal processing, 2010, 90（4):1300-1312.
［14］Nedic A, Ozdaqlar A, Parrilo P A.Constrained consensus and optimization in multi-agent networks［J］.IEEE Transactions on Automatic Control, 2010, 55(4):922-938.
2012, Vol. 33 
