Research On Link Quality Estimation Of Heterogeneous Wireless Sensor Networks

Real-time and accurate link quality estimation is essential for wireless sensor networks.In practical applications,there are many heterogeneous networks in wireless sensor networks,and little researches on link quality estimation of heterogeneous wireless sensor networks.Therefore,the research on link quality estimation of heterogeneous wireless sensor networks is of great value to the design of network protocols.This paper first researches the background,and analyzes the research status at home and abroad of link quality estimation.On this basis,the following contents are completed:(1)Through the measured data,the applicability of the commonly used link quality indicators under homogeneous network in heterogeneous network is analyzed.And the statistical characteristics of link quality indicators applicable to the heterogeneous network are analyzed.(2)Most existing link quality estimators are lack of effective perception and self-adaption to link dynamics.Therfore,a software-based estimator called Fluctuation adaptive Link Quality Estimator(FaLQE)is proposed.The FaLQE adjusts smoothing factor of the estimation according to changing degree of PRR of adjacent estimation windows dynamically and achieves equilibrium of stability and agility accordingly.Compared with existing estimators,the estimate error of FaLQE is smaller.(3)Most existing hardware-based link quality estimators cannot solve the problem of large fluctuations of physical layer parameters in fast estimation.In view of this,a fast link quality estimation scheme based on exponential weighted Kalman filtering is proposed.It obtains more stable estimated values of received signal strength indicator using exponential weighted Kalman filter and then computes signal-to-noise ratios accordingly.Link quality is estimated by the mapping relation between signal-to-noise ratio and packet reception ratio.Experimental results show that,accurate link quality estimation can be obtained quickly only with 16 packets.Compared with existing estimators,the estimate error of the proposed one is smaller.(4)Most existing link quality estimators based on link quality indicatr don't consider the actual physical meaning of mapping relationship between link quality indicator and packet reception ratio.A more accurate link quality estimater is proposed.It obtains more stable link quality indicator estimation values by exponential weighted Kalman filtering.And then,the hyperbolic tangent model is used to estimate the link quality quantitatively,which is hyperbolic tangent model is more in line with the actual physical meaning.Compared with similar estimators,estimate error of the proposed one is smaller.(5)Existing estimators employ too complicated multi-parameter fusion methods,which could not offer a good balance among accuracy,agility and low overhead.The estimator uses the weighted Euclidean distance to fuse link quality indicator and signal-to-noise ratio to obtain a new metric WED.PRR is estimated using mapping relation between WED and PRR,which is fitted by logistic regression.Compared with existing estimators,the estimate error of the proposed one is smaller.(6)The above research can not solve the problem that the estimated value fluctuates greatly and is inaccurate under unstable links.A lightweight and fluctuation insensitive multi-parameter fusion link quality estimator LFI-LQE is proposed.The LFI-LQE uses exponential weighted Kalman filtering to process the physical layer parameters to further eliminate unnecessary fluctuations.Compared with existing estimators,the estimate error of LFI-LQE is smaller.More importantly,compared with other multi-parameter estimators,the computation overhead of LFI-LQE is also greatly reduced.