Research On Dynamic And Realtime Traffic Flow Data Harvesting Correlation Technologies Based On WSN

Real-time traffic flow data, including volume, speed and occupancy, are the typical parameters representing the traffic flow state and characteristics, which constitute the intelligent traffic system as the kernel factors. Till now, the huge capacity, poor quality and regionalization of the traffic flow data harvesting forms the most challenging work. Therefore, based on the wireless sensor network (WSN), this paper solves the problems in traffic flow data harvesting by the catastrophe theory, the second law of thermodynamics, optimal filtering theory, data mining and data aggregate. The main contributions are as follows:(1) A traffic flow data detecting model composed of 3 sensor nodes is proposed on the basis of WSN employing the central network topology, and hierarchical mechanism. The traffic flow data is acquired through the calculation process of the 3th node according to the data sent by the other 2 nodes. Besides, some energy saving strategies are applied on these 3 nodes to extend the model working life. The experimental results proved the superiority of the proposed model to the 2-node detecting model's proposed by UC Berkeley. Furthermore, an optimization algorithm is proposed to adjust the aggregation granularity for the 3-node detecting model based on binomial distribution of traffic flow data. The experimental results demonstrate the effectiveness of the proposed algorithm.(2) An adaptive traffic flow data detecting algorithm is proposed based on Kalman filter and patch estimation theory. According to the 3-node detecting model working process, this algorithm estimate the traffic flow data using Kalman filter, and fit the variance caused by error drift and traffic flow character variation adaptively considering long-term detecting. Simulation results verify the proposed algorithm.(3) The concept of Opportunistic Routing Entropy is proposed. Combining with ant colony algorithm, a new opportunistic routing algorithm is also proposed aiming at dynamic transmission network and manual configuration in the real traffic scene. Referring to the concept of entropy in the second law of thermodynamic, Opportunistic Routing Entropy is defined to reflect the transmission state of each node in the network, which value is inversely proportional to the node energy and directly proportional to the distance. The dynamic establishment of the routing is achieved by improvement of the ant colony optimization algorithm. Experimental results confirm that the routing algorithm is more energy conservative and throughput gainful compared with other existing routing protocols.(4) A distributed data quality analysis algorithm is proposed and applied to wireless sensor network working as the traffic flow data detection platform. Taking advantage of the computing resource on WSN, the algorithm is running online to optimize the data, which contributes to avoiding consuming resources. Based on the Cusp Catastrophe Theory, the evaluation equation is developed. Aiming at self-adaptive and self-adjustable characteristics of the proposed algorithm according to the traffic data detection, the batch estimation filter is adopted based on the active-learning mechanism. The simulation results show that the proposed algorithm outperforms other data quality analysis algorithms with better performance and scalability.(5) To classify an area-wide traffic state with a high-dimensional feature set space, a novel fuzzy clustering algorithm is proposed based on the dimension reduction and Entropy Coding. The algorithm is realized by adding an efficient method based on Entropy Coding to the conventional FCM algorithm, which reduces the dimension of the feature set space and improves the convergence by re-extracting the intermediate results of FCM and renewing the feature sets with the Entropy Coding. Experimental results on real traffic data show the effectiveness and robustness of the proposed algorithm.