【導(dǎo)讀】Email:,MaximumLikelihood(ML)estimator[3],[4],ismoreaccurate

　　

【正文】 inthe centralized scheme, multihop munication is requiredand the data transmission from such sensors still has tofinally reach the reference sensor. On the other hand, in thedistributed scheme, such sensors can obtain the data fromthose who already received the data located within the firsthop from the reference sensor. That requires shorter totalmunication distance and lower energy than what is neededin the centralized scheme. Hence, the advantage of distributedmethod bees more conspicuous when multiple hops areneeded or the number of participating sensors is increases asillustrated Figure 3. We also study the convergence issue forthe distributed method by considering the localization errorat each iteration paring it with the distance differencebetween the consecutive estimates. Figure 4 shows that bothamounts are highly correlated. The advantage of this scenario10 15 20 25 3012x 10?8Number of participating sensorsMSEDistributedCentralizedFig. 1. MSE vs. number of sensors: Distributed method produces smallererror than centralized method.10203040506070809010001x 10?7SNRMSECentralizedDistributedFig. 2. The accuracy of the distributed method is less affected by a lowenergy target signal than the centralized method.is that we can approximately evaluate the current error fromthe sequence of estimates. This is important if we want tosave unnecessary cost when the accuracy has already reachedacceptable or required levels.VII. CONCLUSION AND FUTURE WORKWe proposed a distributed algorithm, based on range difference localization method, which allows time delay estimationto be carried out at each participating sensors. TDOAs puted from time delay estimates are fused using a sequentialleast squares scheme which enables the appropriate sensorselection based on the current estimate. The results illustratethat the distributed localization produces smaller error andconsumes less energy than centralized method. The advantageof distributed processing bees more conspicuous for errorconsiderations when the number of participating sensors is10 15 20 25 30Number of participating sensorsEnergy Consumption (Joule)DistributedCentralizedFig. 3. Energy consumed by centralized method is larger than that consumedin the distributed method.0 5 10 15 20 25012x 10?3Numbers of included sensors DistanceDistance to the actual target locationDifference distance between consecutive estimateFig. 4. Distance error and distance between consecutive estimates are highlycorrelated.small and the better energy saving is obtained when increasingnumber of participating sensors. The proposed method isalso more robust to decreasing target signal energy and theinstantaneous error from the sequence of estimates can beapproximated and used to reconcile the cost and the systemperformance. In the future we aim to study the effect of timesynchronization errors on time delay estimation, and thus, thelocalization performance.REFERENCES[1] J. C. Chen, L. Yip, J. Elson, H. Wang, D. Maniezzo, R. E. Hudson,K. Yao, and D. Estrin, “Coherent acoustic array processing and localization on wireless sensor works,” in Proceedings of the IEEE, vol. 91,Aug 2020.[2] Q. Wang, W. Chen, R. Zheng, K. Lee, and L. Sha, “Acoustic targettracking using tiny wireless sensor devices,” in IPSN 2020, 2020.[3] J. C. Chen, R. E. Hudson, and K. Yao, “Maximumlikelihood sourcelocalization and unknown sensor location estimation for widebandsignals in the nearfield,” IEEE Transactions on Signal Processing, 2020.[4] R. J. Kozick and B. M. Sadler, “Source localization with distributedsensor arrays and partial spatial coherence,” IEEE Transactions onSignal Processing, 2020.[5] X. Sheng and Y. Hu, “Energy based acoustic source localization,” inIPSN, 2020.[6] W. R. Heinzelman, A. Chandrakasan, and H. Balakrishnan, “Energyefficient munication protocol for wireless microsensor works,” inProc. 33rd Hawaii International Conference on System Sciences, 2020.[7] S. Phoba, N. Jacobson, and R. Brooks, “Sensor work based localization and target tracking through hybridization in the operational domainsof beamforming and dynamic spacetime clustering,” in GLOBECOM2020, 2020.[8] M. Chu, H. Haussecker, and F. Zhao, “Scalable informationdrivensensor querying and routing for ad hoc heterogeneous sensor works,”International Journal of High Performance Computing Applications,2020.[9] W. Chen, J. C. Hou, and L. Sha, “Dynamic clustering for acoustictarget tracking in wireless sensor works,” IEEE Trans. on MobileComputing, vol. 3, no. 3, JulSep 2020.[10] C. H. Knapp and G. C. Carter, “Time delay estimation,” in ICASSP’76,1976.[11] Y. Huang, J. Benesty, and G. W. Elko, “Realtime passive sourcelocalization: A practical linearcorrection leastsquares approach,” IEEETrans. Speech and Audio Processing, vol. 9, no. 8, Nov 2020.[12] Y. T. Chan and K. C. Ho, “A simple and efficient estimator for hyperboliclocation,” IEEE Transactions on Signal Processing, 1994.[13] G. C. Carter, “Time delay estimation for passive sonar signal processing,” IEEE Transactions on Acoustics, Speech, and Signal Processing,1981.[14] S. M. Kay, Fundamentals of Statistical Signal Processing. PrenticeHall, 1993.[15] T. H. Cormen, C. E. Leiserson, R. L. Rivest, and C. Stein, Introductionto Algorithms. McGrawHill, 2020.