Thereafter, FTTT is extended by quantifying the pairwise uncertainty to conduct sampling vectors.
* Extensive simulations and an outdoor experimental system are given to validate the correctness and efficiency of FTTT strategy.
Section 4 introduced the Fault-Tolerant Target-Tracking strategy (FTTT) in detail and Section 5 presents its performance by analyzing the proper sampling times and the overall tracking error.
In this section, the Fault-Tolerant Target Tracking (FTTT) strategy will be detailed discussed.
An adaptive grid division algorithm in our previous work  can be used to simplify the face division pre-process of FTTT. However, the detailed discussion is beyond the capability of this paper.
We can see that FTTT is based on the grouping sampling which means each sensor should sense several times in a very short time interval.
One is caused by the distance between the located face of FTTT and the actual face that the target located in, which is called inter-face error.
We conduct simulation experiments to compare the proposed Fault-Tolerant Target-Tracking (FTTT) strategy with Direct maximum likelihood estimation (Direct MLE)  tracking scheme and the optimal path matching strategy with MLE (PM) proposed in .
This subsection gives an intuitive comparison between FTTT and PM strategy of a tracking example.
11(a) shows the dynamic tracking errors of FTTT, PM and Direct MLE along with time series, which further proves the conclusion that the tracking performance of FTTT is much greater than the other two.
1) Tracking Performance Comparison with Different Number of Sensor Nodes: We compare the FTTT strategy with PM and Direct MLE methods under different number of random deployed sensor nodes, which varies from 5 to 40.
12(a) shows the mean tracking error of FTTT changes with the sensing resolution [epsilon] when randomly deploying 10,15,20 and 25 sensor nodes.