Studies On Energy-Efficient Algorithms For Data Collection In Wireless Sensor Networks

Wireless sensor networks are composed of a large number of sensor nodes self-organizing into networks, and the major task is environmental data collection, involving wireless communication, data processing, routing and many other technologies. Due to limited computing, storage, communication and energy resources of nodes, how to make full use of limited resources to collect data energy-efficiently is of great importance. The paper studies the problem, and begins with data-aggregation based routing, mobile sink, wavelet-based data processing, work node selection, etc. to investigate the ways of energy-efficient data collection for wireless senor networks. The main research results obtained are as follows:1. Data aggregation can reduce the in-network data transmission, but with non-full data aggregation, the nodes transmitting aggregated data bear heavy loads, dying prematurely. For this reason, an adaptive state-aware routing algorithm for data aggregation is proposed. Initially, the algorithm builds a hop-tree based on nodes’state. Then, for an event, the algorithm clusters nodes within event areas in a distributed manner, constructs the shortest path with the best state from the coordinator to the hop-tree backbone or the sink according to the residual energy of nodes, and utilizes the TTL to restrict the update range of the hop-tree, which realizes the moderate overlap between event paths. Algorithm analysis and experiments show that the algorithm can improve data aggregation efficiency and balance energy consumption, realizing energy-efficient data collection.2. There is always the "hot-spots" problem in multi-hop wireless sensor networks with immobile nodes. In order to solve this problem, an energy efficient algorithm based on cluster and hop-tree for single mobile sink is proposed here. After nodes within event areas clustering in a distributed manner, the algorithm calculates a new quasi-optimal position based on cluster information for the mobile sink, and determines a shorter moving route to the position according to the obstacle information. As the mobile sink arrives at the position, the algorithm updates the routing structure, constructs a hop-tree based on event clusters, aggregates intra-cluster data and performs reliable data transmission. Algorithm analysis and experiments show that the algorithm can shorten data transmission distance effectively while balancing node energy consumption, move the mobile sink efficiently, cost reasonable control overheads, and collect data energy-efficiently.3. Although routing for improving data aggratation efficiency can make data be aggregated as soon as possible, it may bring about the unbalanced network load and even longer data transmission distance. If a mobile aided sink is employed to help the immobile primary sink, it makes for improving data aggregation efficiency, and decreasing and balancing network loads. Based on the above considerations, a data-aggregation and dual-sinks (an immobile primary sink and a mobile aided sink) based routing algorithm is proposed. The algorithm utilizes neighbor information comparison and delayed sending of control messages to decrease distributed clustering overhead, gets the routing aggregation center according to cluster information and selects the neighbor nearest to the primary sink and the routing aggregation center as the next hop for each node, building an approximate Steiner tree with the primary sink as its root. Using the routing aggregation center, the algorithm drives the aided sink move to the neighborhood of the paths bearing heavier loads to intercept aggregated data. Algorithm analysis and experiments show that the algorithm can aggregate in-network data as soon as possible, and decrease and balance energy consumption.4. Haar wavelet could be used by wireless sensor networks to reduce redundant data within the network, saving energy spent on data transmission. Available algorithms fail to optimize the data compression performance for Haar wavelet from the perspective of data, so a data preprocessing algorithm for better Haar wavelet based data compression is proposed. The algorithm utilizes the sample means of sensed data to adjust the intra-cluster node order list, smoothes the data to be processed, restricts the frequency of the update on node sample means and intra-cluster node order list by some defined thresholds, making the adjustment of data smoothness not too often, which saves control overheads. Algorithm analysis and experiments show that the algorithm can promote the performance of Haar wavelet based data compression:improve the accuracy of reconstructed data and cost reasonably, or increase the degree of data compression, saving energy spent on data transmission.5. How to extract the most representative key nodes within an event area is very valuable for both the access of key information and the decrement of data transmission. Inspired by morphological erosion and dilation operations on binary images, a distributed and morphological operation based data collection algorithm is proposed. In the algorithm, a square structure element is adopted and each node gets the structure neighbors based on it. According to the event monitoring situation of its structure neighbors, each node decides whether it should be eroded autonomously, then the backbone of the event area can be acquired and only these backbone nodes are allowed to send the their sensory data. The sink obtains the backbones of event areas according to the data collection situation, and recovers the event area by the dilation operation to direct the offline work. Algorithm analysis and experiments show that the algorithm can extract the backbone area which retains the original shape of an event area in a quick and low overhead manner, decrease the amount of data transmission, gain key data, and estimate the approximate ranges of event areas.