An Energy Consumption Optimized Clustering Algorithm for Radar Sensor Networks Based on an Ant Colony Algorithm

Radar sensor networks (RSN) [1] are the networks of distributed radar sensors which collaboratively operate and are deployed ubiquitously on airborne, surface, and unmanned vehicles in a large geographical area. Radar sensors have capabilities for radar sensing, signal processing, and wireless communications. In RSN, radar sensors are networked together in an ad-hoc fashion, that is, they do not depend on any preexisting infrastructure. In fact, they are self-organizing entities that are deployed on demand to perform various tasks such as surveillance, search and rescue, disaster relief, and so forth.

An RSN is organized into clusters, which are independently controlled and dynamically reconfigured to observe targets such as tactical weapons, missiles, aircraft, ships, and so forth, in the surveillance area. In a cluster, sensors receive the signals backscattered by targets in the presence of interference (e.g., clutter, jamming, interference between radar sensors) and noise. Then, the observed signals from all radar sensors are forwarded to a cluster head where received data set will be combined to perform fundamental tasks such as detection, localization, identification, classification, and tracking. After finished dealing with the data from the radar sensors, the cluster head transmits the data to the sink node. In RSN, the cluster head must have a function such as data collection, fusion, transition, and so forth. The cluster head consumes more energy than the radar nodes. Hence, the algorithm of selecting the cluster head is very important for the life cycle of the radar sensor network.

The routing for wireless sensor network [2] (WSN) based on bionics heuristic algorithm has become one of the hot WSN research topics. Currently, ant colony algorithm [3], EEABR [4], MACS [5], and EPACOR [6] are utilized to find the best path from source to sink for wireless sensor network. In addition, ant colony optimization (ACO) [7], EAQR [8], QDV [9] not only solves the WSN network congestion, but also improves QOS and network security. Since the clustering algorithm has a good performance on the WSN routing, Salehpour et al. [10] proposed an efficient routing algorithm for large-scale WSN. This algorithm takes advantage of two routing levels. In the first level (intracluster), cluster members send data directly to their cluster head and the cluster head is selected by LEACH [11]. In the second level (inter-cluster), the cluster head uses ACO algorithm to find a route to the base station. ACDCHA [12] (Ant Colony-based Double Cluster-Heads Algorithm) is proposed by Su et al. through introducing the idea of ant colony algorithm. A master cluster head and a vice cluster head are selected in each cluster according to the pheromone concentration. The tasks such as data collection, fusion, transition, and so forth, are allocated, respectively, to these two kinds of cluster head. Simulation results show that this algorithm can get a better balance of the energy dissipation and prolong the network lifetime than the traditional clustering algorithm. The benefit of this algorithm is that the master cluster head only sends the data to the vice cluster head within the cluster, then the vice cluster head sends the data to base station, thus saving the energy of the master cluster head. However, when the master cluster head and vice cluster head are selected, only the residual energy of the node is considered; the node's location information is not taken into account, so the distance between vice-cluster head and base station maybe greater than the distance between the base station and the master cluster-head, resulting in more energy consumption for vice cluster head.

Heinzelman proposes a "Low-Energy Adaptive Clustering Hierarchy" (LEACH) for WSN data gathering and communication [11]. In this protocol, the base station is assumed to be far away from the area where sensors are deployed. The WSN is clustered according to a certain rule and the cluster head collects data from all the sensor nodes within its cluster. Once the data from each sensor node are received, the cluster head aggregates and transfers them directly to the base station, meaning that being a cluster head cost much more energy than regular sensors. In order to make all the sensor nodes use up their energy approximately simultaneously, the role as the cluster head is randomly rotated among all sensor nodes in the same cluster. From this point of view, LEACH is a quite elegant protocol especially in terms of self-adaptive ability. But there are many disadvantages in traditional clustering routing algorithm LEACH. First of all, the cluster head is selected by probability model, so it possibly causes some cluster heads apart from each other or some nodes in part areas apart form cluster head, then transmission consumption of these nodes increases. Meanwhile, LEACH does not consider the residual energy of each node, so the nodes that have relatively small residual energy can be the cluster heads. This makes the network lifetime to be shortened. Secondly, the randomness of cluster head selection can cause that some cluster heads share burden of nodes unbalance and aggravate some cluster head node's burden. So the balance degree gets reduced. For these deficiencies, ESCAL [13] and NRSM [14] have improved this algorithm and effectively extend the network lifetime.

To overcome the defect of LEACH and ACDCHA, in this paper, the cost function for clusters is proposed based on radio-free space path loss for radar sensor network. The radar node joins the different clusters according to the values of cost function between this node and its cluster head, so the cluster structure is optimized. In the cluster head selection phase, the energy and distance pheromones based on the residual energy and aggregation of radar sensor nodes are proposed. Based on ant colony algorithm, we propose a new ant colony-based clustering algorithm (BACCA) for radar sensor network. Simulation results show that this algorithm can keep a better balance of the energy consumption and prolong the life cycle of the radar sensor network.

The rest of this paper is organized as follows. Several concepts such as the cost function for clusters, energy pheromones, distance pheromones, and cluster head selection function are introduced in Section 2. The BACCA is described in Section 3. Simulation model and numerical results are discussed in Section 4. Finally, the conclusions are drawn in Section 5.

According to the above analysis, we are ready to get the BACCA algorithm for radar sensor network as follows:

Clearly, in this algorithm, when the new cluster head is chosen, both the residual energy and aggregation of radar nodes are taken into account. At the same time, when the radar nodes choose to join the corresponding cluster head to form the new cluster, both the residual energy of radar node and energy attenuation of data transmission are considered. Therefore, by using this algorithm, the radar sensor network lifetime can be prolonged effectively.

In this paper, the ant colony algorithm is applied to RSN clustering routing. The energy and distance pheromones which are based on the residual energy and aggregation of radar sensor nodes are proposed. We also propose the ant colony-based clustering algorithm (BACCA) for radar sensor networks according to the ant colony algorithm. Meanwhile, the cluster structure is optimized by the defined cost function for clusters. The simulation results show that this algorithm is better than LEACH and ACDCHA in that it can better balance the energy dissipation of the radar sensor network and extend the lifetime of the radar sensor network.