A Prediction-Based Location Update Algorithm in Wireless Mobile Ad-Hoc Networks

Location information in Mobile Ad Hoc Networks (MANETs) provides efficiency to uni-cast and multi-cast algorithms. This paper proposes a new location update algorithm called PLU. The essence of PLU lies in the integration of location prediction and one-hop broadcasting of location update packets. The full PLU algorithm and its associated data structure are described in the paper. The simulation results have shown an improved performance of PLU thanks to its newly introduced type of updates where location prediction is applied to reduce the number of packets or bytes transmitted for location update purpose whereas keeping a high accuracy level of location information. 1 Background and Motivation Mobile ad-hoc networks (MANETs) consist of a set of wireless mobile nodes that cooperatively form a network without a fixed infrastructure. In such a wireless network, a message sent by a node usually reaches all its neighbouring nodes that are located within the transmission radius of the sender. Because of the limited transmission radius, the routes between the original sender node and the intended final receiver node normally consist of several hops. As such, each node in a MANET also serves as a router to route information between neighbours, thus contributing to the end-to-end message delivery across the whole network. Routing plays a critical part in the practical success of any MANET. Many routing protocols have been proposed for MANETs, and a comprehensive review of these protocols can be found in [1]. Recently, more and more researchers in MANET community realized the importance of location information of nodes in MANET routing and some locationaided routing algorithms were put forward, such as Location-Aided Routing (LRA) [2], the Distance Routing Effect Algorithms for Mobility (DREAM) [3], and the Geographical Routing Algorithm (GRA), amongst others. I. Stojmenovic gave a wellsummarized survey of most of the typical location-based routing algorithms in ad-hoc networks [4]. It is believed that the advantages of using location information outweigh the additional cost [5]. This is further justified by the increasing availability of small, inexpensive low-power GPS receivers and techniques for finding relative coordinates based on signal strengths. Given the fact that location information of nodes can be obtained using whatever way, the next step is how to utilize them effectively and A Prediction-Based Location Update Algorithm 693 efficiently to benefit the routing in MANETs. This is typically what the location information service (LIS) is supposed to do. Actually knowing other nodes’ location information is also useful, and sometimes vital, in some cases other than routing, such as life rescue, movement decision making in war field, etc. The essence of a location information service is its location update scheme. The ultimate goal of any LIS is to reduce the number of location update packets (LUP) or bytes (as overhead) transmitted across the MANET for the maintenance of LIS whereas keeping as high as possible the accuracy level of location information. To this end, many location update schemas are proposed in the current literature. The simplest location update scheme is location information flooding where each node broadcasts its own location information in a flooding way on a periodic basis. A location table, containing nodes’ location information received by the node, is maintained in every node in the network. Flooding causes massive transmission of redundant messages and consequently collisions, and usually is not an ideal solution for location information updating. So a number of improved location information service algorithms were proposed. The location information service utilized in the DREAM routing protocol [3], referred to as DREAM Location Service (DLS) here, takes into consideration the distance factor when sending location updating packets. In DLS, if the distance of two nodes is further away then less updates are produced – this is because faraway nodes appear to move more slowly than nearby nodes. DLS classifies the whole nodes in a network into two types: nearby nodes and faraway nodes. Each mobile node in the MANET transmits an LUP to nearby nodes at a given rate and to faraway nodes at another lower rate. By reducing the frequency of sending location updating packets to faraway nodes, the overall DLS overhead is reduced. T. Camp et al. discussed in [6] another LIS algorithm named Simple Location Service (SLS). SLS also transmits LUP to its neighbours at a give rate. The difference between SLS and DLS lies in the type of information exchanged and the distance the information is propagated. Specifically, SLS transmits table containing multiple nodes’ locations to neighbours and then neighbours will carry out some processing before the next hop transmission; whereas DLS transmits only the sending node’s location to its neighbours and then immediately to other nodes via neighbour. Our LIS algorithm proposed in this paper, called PLU (short for Prediction-based Location Updating), also utilizes the table transmission idea as that in SLS but is also different in terms of the content of the table entry, updating strategy, etc. In the same paper [6], T. Camp et al. also presented another LIS algorithm called Reactive Location Service (RLS). RLS is similar to LAR [2] but has a comprehensive mechanism to support location inquiry. In some research work, nodes as location servers were proposed to maintain location information of some other nodes in a MANET. A typical example of this kind is Grid Location Service (GLS). One common aspect of all these abovementioned algorithms is that none of them beard the idea of prediction. A proper prediction of node’s current location based on its previous location and movement model has the potential to significantly reduce the number of location update packets (and the computation is far more energy-economic than transmission). [7] realized the importance of location prediction in location information system and routing and applied it for efficient QoS routing in MANETs. Later on, similar location prediction mechanism appeared in Agarwal, et al’s work 694 J. Shen, K. Yang, and S. Zhong [8]. However, we think the prediction effort made in [8] was undermined to some extent by its adoption of a simple flooding algorithm for location information updating. Inspired by the prediction effort made in [7] and [8], this paper exploits the integration of location prediction with a one-hop broadcasting algorithm. The proper cooperation and better-off balance between location prediction and location update constitute one of the important investigation of this paper. Based on the above discussion, Section 2 details our location information service algorithm PLU, which is followed by simulation environment and parameter design in Section 3 and simulation result analysis and discussion in Section 4. The last section, Section 5, concludes the paper.