Large-scale geographic environmental monitoring and physical infrastructure for transferring and processing data cannot maintain the same growth rate with the monitoring scale, thus make the data gathering and processing under unreliable links, showing a kind of saturation flow state; and the ability of wireless sensor network (WSN) seemingly cannot be stabilized. On the other hand, though the calculation results from ideal network model are accurate enough, they deviate from the practical application greatly, hence the network users cannot adequately analyze and utilize the sensed data from industrial field network, and also cannot analyze the influence of network size and performance on data fusaggregation. Because of these, we will present the ‘transitional region phenomenon’ to be one of assumptions for industrial field simulation, and propose a data fusaggregation algorithm for the practical application on this basis, i.e. data fusaggregation algorithm based on reliability (R algorithm). When designing the R algorithm, as an example of sum operator, the lower limitη of reliability of appreciate aggregation sum result will be calculated by analyzing and synthesizing the result automatically. Then the aggregation sum result and the value ofη will be sent to users together. In addition to providing the summary information from the monitored area to users, R algorithm also provides the parameterηas the judgment of information reliability to facilitate users to do further handling of aggregation results and improve the WSN sensing performance. Simulation results describe the changing rule of reliabilityηcaused by unreliable links from the signal-to-noise ratio in transitional region, and discuss the network influence of size and performance on reliability η, with the increase of network operation cycles and network scale, when the value of reliability η becomes gradually close to 0. And this provides theoretical foundations and empirical formulas for WSNs from theoretical model to practical industrial application.%大规模的地理环境监测，以及用来传输与处理数据的物理基础设施无法和监测区域的规模保持同样的增速，使得不可靠链路下的数据采集与处理呈现出一种饱和流状态，无线传感器网络的能力看似难以稳定。另一方面，尽管理想网络模型的计算结果足够精确，然而，由于与实际应用偏差甚大，使得网络用户无法充分地分析和利用从工业现场所获得到的网络感知数据，并且没有针对网络规模和性能对数据聚融的影响进行分析。为此，本文提出以“过渡区”作为工业现场仿真的假设条件，并在此基础上提出了一种面向实际应用的数据聚融算法，即基于可信度的数据聚融算法(R算法)。在具体设计R算法过程中，选用聚集和操作符SUM为例，通过对网络提供出的近似聚集和加以自动分析、综合，针对相对误差界限ε，计算出近似聚集和的可信度的下限η；并将近似聚集和、参数η一同提供给用户，在为用户提供网络概要信息的同时，还提供了参数η作为对信息可信度的判断，以便指导用户对数据聚集结果进行深度处理和提高网络的感知性能。仿真实验描述了过渡区内由于信噪比导致的链路不可靠所引起的η的变化规律；讨论了网络性能和规模对η的影响，随着网络运行周期的增加和网络规模的增大，η的值将逐渐靠近0；从而为WSNs从理论模型投入实际工业应用提供了理论依据和经验公式。
展开▼
