Lifetime improvement of wireless sensor network by information sensitive aggregation method for railway condition monitoring

Abstract

Lifetime maximization is an important issue while designing wireless sensor network. One of the ways to increase the lifetime of WSN is to reduce the energy demand of the sensor nodes and cluster head nodes, which can be decreased by filtering out redundant data traffic. A Two-Layer Hierarchal Aggregation (TLHA) protocol has been proposed, which filter out redundant data traffic. Simple aggregation algorithm has been used to classify the sensed data based on their normalized standard deviation for efficient data filtration. The algorithm also performs k-Means based re-classification of data to improve the efficiency of the algorithm. Two-layered redundant data filtration technique has been proposed to classify the data. Aggregation protocol reduces data traffic in lower layer hierarchy (sensor nodes to cluster head node) as well as upper layer hierarchy (cluster head nodes to base station node). In former, Energy efficient Time Division Multiple Access (EA-TDMA) Medium Access Control (MAC) protocol is used while in latter Bit-Map-Assisted (BMA) MAC protocol is used for transmission of data. Simulation and experimental results show that TLHA saves an enormous amount of energy, which finally increases the lifetime of the sensor network. The performance of the proposed TLHA protocol has been compared with already existing protocols viz. Data Aggregation Window Function (DAWF) and Spatial-Temporal Correlation Algorithm (STCA).

Introduction

LIFETIME and energy demand of Sensor Node (SN) are two inter-related crucial factors of Wireless Sensor Network (WSN) [1]. Lifetime and robustness of the sensor network can be improved by reducing/removing redundant data transmission [2]. Generally, the sensor nodes (SNs) are small, cheaper, require low processing power and have limited power supply, due to which address-centric operations are irrelevant. Therefore, SNs operate in data-centric paradigm. Such SNs are deployed redundantly to protect the network from the failure of nodes. In a data-centric communication, sensed value is important rather than its source. Due to the low processing capacity of the SN, simple algorithms are used for data aggregation e.g. averaging or mean [3]. Data aggregation is an important method by which data coming from various SNs are aggregated at the Cluster Head (CH) node and the same has been transmitted to the upper hierarchal network [4]. Data aggregation is necessary in most of the cases for better utilization of allotted channel. When data traffic reduces by aggregation, the energy demand of the SNs and CH node also reduces, which improves the lifetime of SN and CH node as well as network. Network lifetime has been defined in various ways by different researchers [3], [5], [6]. Generally, it is the time for which network fulfills its tasks. Various works have been reported by the researchers on data aggregation, which are divided into various classes e.g. cluster-based data aggregation, tree-based data aggregation etc. [7], [8], [9], [10], [11]. In the present work, our focus is tree-based data aggregation method, in which parent node collects the data from child nodes and the same is further transmitted to the upper hierarchal parent.

This work concentrates on WSN based Railway Track Condition Monitoring Application (RTCMA) [12], [13], [14], [15], [16], [17], [18], [19]. In RTCMA, both continuous monitoring and event-driven sensors are present on railway track [20]. Continuous monitoring is often more expensive and has high-energy demand. Periodic or event-driven monitoring is cheaper and consumes less energy. The problem of data redundancy originates in such systems due to the presence of a large number of continuous monitoring sensors e.g. temperature, acceleration, pressure sensor etc., placed on each track periodically to measure similar parameters from different SNs. On each track, single SN is connected with a large number of sensors through General-Purpose Input Output (GPIO) pins. The RTCMA system has been reported in [19], in which three SNs from three nearby tracks transmit their data to the CH node. SN works as a child node and CH node works as a parent node representing a simple star topology-based single-hop network. All the data collected by the CH node in Lower Layer Hierarchy (LLH) are processed for Upper Layer Hierarchy (ULH), where CH nodes transmit collective data to the Base Station (BS) node by single-hop communication. The architecture of RTCMA is shown in Fig. 2(a). Data collected at CH node comprises some redundant data, due to which, there is wastage of bandwidth as well as energy. All the data collected by the CH node in Lower Layer Hierarchy (LLH) are processed for Upper Layer Hierarchy (ULH), where CH nodes transmit collective data to the Base Station (BS) node. In this paper, an efficient aggregation protocol named Two-Layer Hierarchal Aggregation (TLHA) has been proposed that filters redundant data traffic. The proposed network has been implemented with two MAC protocols. The two MAC protocols are Energy efficient Time division multiple Access (EA-TDMA) and Bit-Map-Assisted (BMA). The former is used in LLH and the latter is used for ULH. EA-TDMA and BMA operation is divided into two phases, the set-up phase, and the steady-state phase. The former is cluster formation phase while the latter is further divided into two parts i.e. Contention Period (CP) and Data Transmission Period (DTP). Operational diagram of EA-TDMA and BMA is shown in Fig. 1(a) and (b) respectively. Shaifullah et al. had already proposed EA-TDMA [21] and E-BMA energy-efficient MAC protocols for Railway Monitoring Applications (RMA) [22]. In the present work, our focus is to maximize the lifetime of WSN using data aggregation without compromising the quality (accuracy) of the data. Therefore, the data aggregation operation has been divided into two layers i.e. LLH and ULH. In the former, temporal aggregation operation is performed where SN compares current reading of sensed data with the previous one and decides whether to transmit the data or not. In the ULH, CH nodes collect the data and classify the collected data into different classes, based on their deviation w.r.t. mean value, to perform aggregation operation. This classification helps to filter redundant data traffic. Performance of TLHA has been analyzed in terms of accuracy and energy efficiency of the network. In the present work, the performance of proposed algorithm has been analyzed for railway monitoring application but the algorithm can be used for other applications as well e.g. WSN for agriculture field, rescue operations in the military etc.

The novelty of the present work is summarized as follows:

• The proposed aggregation algorithm performs two-layered temporal and spatial aggregation operation with three classification windows.

• The proposed aggregation scheme also performs k-Means based re-classification of data to improve the aggregation efficiency.

• The proposed aggregation algorithm is implemented with the help of EA-TDMA and BMA MAC protocol.

The rest of the paper has been presented as follows. Section 2 describes the related works. Section 3 describes the problem statement and its possible solutions. Section 4 describes the proposed aggregation protocol. Section 5 describes the analytical modeling for energy consumption and lifetime of WSN. Section 6 discusses the simulation results and performance analysis. Section 7 discusses the experimental results and data analysis. Section 8 concludes the major research findings of the proposed work.