A Simulation Model and a Hybrid Genetic Algorithm for Energy-Aware MANET Routing and Planning Ivana Cardial de Miranda Pereira, M.Sc Centro de Análise de Sistemas Navais (BrazNavy) COPPE, Federal University of Rio de Janeiro, Brazil ivanacardial@hotmail.com Nelson Francisco Favilla Ebecken, D.Sc. COPPE, Federal University of Rio de Janeiro, Brazil nelson@ntt.ufrj.br ABSTRACT This paper presents a new model developed to aid the planning and the analysis of communications-intensive Mobile Ad Hoc Networks (MANET), with respect to the allocation of energy- critical equipment. A graphical simulation tool and a new hybrid genetic algorithm (HGA) are introduced. They work together to estimate the required amount of deployed battery supplies and the probability of success of real operations. At each period, a hybrid genetic algorithm with reparation of individuals and heuristic crossover and mutation operators finds efficient routes that preserve maximum energy availability at network level, reducing the probability of communications disruption. The simulation tool implements mobility models derived from experts’ advices and may be used in missions like military and search-and-rescue operations. One may easily include new models to represent the movement of nodes in other specific missions, including trace data. The system is flexible and customizable, providing a means to mission planning, including the provision of adequate power supply for the large number of devices typically included within a MANET. General Terms Wireless ad hoc networks. Keywords Genetic Algorithms, Simulation, MANET, Energy Efficiency. 1. INTRODUCTION A mobile ad hoc network (MANET) is a network that has many free or autonomous nodes, composed of mobile devices that can dispose themselves in various ways and operate without strict top-down, administration, or a fixed infrastructure. Each node must forward traffic not necessarily related to its own use, actually acting as a router. The primary challenge in building a MANET is equipping each node to unceasingly keep the information required to correctly route traffic. The development of small smart communications devices, laptops and 802.11/Wi-Fi wireless networking has made MANETs a popular research topic since the mid-1990s. Presently, the ever- growing use of wireless communication technology is raising an import issue: the restrictions imposed by the lifetime of the devices’ batteries. MANETs vary from small groups with restricted mobility and low energy requirements, as interactive sessions in universities, to large, highly mobile networks with a dynamic topology and long duration, where energy consumption becomes an important issue. The design of routing protocols for this kind of network has always been a highly complex task. Due to the unforeseen mobility of the nodes, ad hoc networks present a dynamic topology, i.e., their geometry change frequently and unpredictably, stressing the need for effective and efficient routing protocols. The discovery of viable routes and the efficient delivery of data packets in a decentralized and continuously changing environment is an ill-defined problem. Some relevant technical and topological aspects - inconstant QoS (quality of service), propagation losses, mutual interference, energy consumption, and a highly mobile setup of the entire network - are extremely relevant in this context. Because of that, MANETs must be able to update its routes in an adaptive fashion. In military operations, for example, issues like security, latency, reliability, intentional interference or jamming, and recovery from failure must be seriously considered [1]. These applications typically need coordinated actions, with traffic patterns that obey a rigid hierarchical chain of command, and a "controlled" randomness of the movement of nodes, following a pre-determined pattern of mobility, since the groups must pursue, in a collaborative way, common objectives [2]. This work presents new models and computer tools designed to help the planning process of MANETs employed in military and search and rescue operations, or other operations with similar characteristics, like those deployed in inhospitable and/or energy-critical environments. A new hybrid genetic algorithm (HGA) solves the routing problem, aiming to optimize the total energy load of the network, as well as trying to minimize network disruption caused by battery exhaustion by one or more nodes. The main objective here is to increase the probability of mission success. A simulation model to position objectives and moving groups of wireless stations (people, vehicles, etc.) in a limited area of operations is implemented. Beyond producing a graphical presentation of the kinematics of the whole operation, the simulator provides input data for the HGA. Several parameters may be determined to establish the geographical environment scenario, to position stations inside each group, to model each station´s mobility on their way to the objective as well as the characteristics of the communication devices, as range and power consumption, etc. 2. MOBILE AD HOC NETWORK 2.1 General Description Wireless communications networks may be classified in two principal classes: infrastructured and ad hoc networks. In the first kind, all communications between nodes occur through fixed support stations. Nodes do not communicate directly with each other, even if they are within range. An ad hoc network, as shown in Figure 1, is composed of a set of independent nodes with wireless communication capability. They communicate directly among themselves, dynamically forming a temporary network, without the need for any central access point or fixed support station. Each node, beyond its role as end-user station running applications, works as a router, and is able to discover and maintain a set of possible routes to other nodes.