دیدگاه ترکیبی از تحرک درشبکه های موردی سیار :مدل های تحلیلی و مطالعه شبیه سازی

We study the effects of node mobility on the wireless links and protocol performance in mobile ad hoc networks (MANETs). First we examine the behavior of links through an analytical framework and develop statistical models to accurately characterize the distribution of lifetime of such wireless links in MANETs. We compute the lifetimes of links through a two-state Markov model, and use these results to model multi-hop paths and topology changes. We show that the analytical solution follows closely the results obtained through discrete-event simulations for two mobility models, namely, random direction and random waypoint mobility models. Finally, we present a comprehensive simulation study that combines the results from the findings in simulations with the analytical results to bring further insight on how different types of mobility translate into protocol performance.

The communication protocols of mobile ad hoc networks (MANET) must cope with frequent changes in topology due to node mobility and the characteristics of radio channels. From the standpoint of medium access control (MAC) and routing, node mobility and changes in the state of radio channels translate into changes in the state of the wireless links established among nodes, where typically a wireless link is assumed to exist when two nodes are able to decode each other’s transmissions. The behavior of wireless links is critical to the performance of MAC and routing protocols operating in a MANET. However, no analytical model exists today that accurately characterizes the lifetime of wireless links, the paths they form from sources to destinations, and topology changes resulting from dynamic link behaviors, as a function of node mobility. As a result, the performance of MAC and routing protocols in MANETs have been analyzed through simulations, and analytical modeling of channel access and routing protocols for MANETs have not accounted for the temporal nature of MANET links and paths. For example, the few analytical models that have been developed for channel access protocols operating in multi-hop ad hoc networks have either assumed static topologies (e.g., [1]) or focused on the immediate neighborhood of a node, such that nodes remain neighbors for the duration of their exchanges (e.g., [2]). Similarly, most studies of routing protocol performance have relied exclusively on simulations, or had to use limited models of link availability (e.g., [3]) to address the dynamics of paths impacting routing protocols (e.g., [4]). Accordingly, there is a strong motivation to investigate analytical modeling of link dynamics and its generalization to the distribution of multi-hop paths and topology changes. This paper provides the most accurate analytical model of link and path behavior in MANETs to date, and characterizes the behavior of links and paths as a function of node mobility. The importance of this model is twofold. First, it enables the investigation of many questions regarding fundamental tradeoffs in throughput, delay and storage requirements in MANETs, as well as the relationship between many crosslayer-design choices (e.g., information packet length) and network dynamics (e.g., how long links last in a MANET). Second, it enables the development of new analytical models for channel access, clustering and routing schemes by allowing such models to use link lifetime expressions that are accurate with respect to simulations based on widely used mobility models. Recently, Samar and Wicker [5] and [6] pioneered the analytical evaluation of link dynamics, and provided new insight on the importance of an analytical formulation of link dynamics in the optimization of the protocol design. However, Samar and Wicker assumed that communicating nodes maintain constant speed and direction in order to evaluate the distribution of link lifetime. This simplification overlooks the case in which either one of the communicating nodes changes its speed or direction while the nodes are in transmission range of each other. Consequently, the results predicted by Samar and Wicker’s model could deviate from reality greatly, being overly conservative and underestimating the distribution of link lifetime [5] and [6], especially when the ratio R/v between the radius of the communication range R to the node speed v becomes large, such that nodes are likely to change their velocity and direction during an exchange. The first contribution of this paper is to provide a two-state Markov model that better describes the mobility patterns of communicating nodes. Section 2 describes the network and mobility models used to characterize link and path behavior. Section 3 describes the proposed analytical framework and presents our results on link lifetime. Our approach is based on a two-state Markovian model that reflects the movements of nodes inside the circle of transmission range and builds an analytical framework to accurately evaluate the distribution of link lifetime. And the proposed model subsumes the model of Samar and Wicker [5] and [6] as a special case, and provides a more accurate characterization of the statistics of link lifetime. Section 5 illustrates the accuracy of our analytical model by comparing the analytical results against simulations based on the random direction mobility model (RDMM) and the random waypoint mobility model (RWMM) model. We extend the analytical results on link dynamics to path dynamics and topology changes in Section 4. These results help to understand the difference in protocol performance obtained in the subsequent simulation study. We argue that, although the end-to-end performance measure such as delay and packet delivery ratio are important, a complementary view of the performance of MAC layer is also essential to obtain a thorough understanding on how mobility influences protocol performance. Although a simulative framework is presented in [7] to incorporate mobility effects into simulations of ad hoc networks, little works has been reported in analyzing the performance evolutions of MAC layer due to nodes’ mobility. Section 6 presents a comprehensive simulation study on the mobility induced performance evolution of the MAC layer. To the best of our knowledge, it is the first work to provide such results. Section 7 concludes this paper.

We focused on the problem of understanding the mobility effect on the protocol performance from a hybrid view, both analytically and simulated. We presented an analytical framework for the characterization of link and extend it to describe path lifetime and topology stability in MANETs. Given the existence of prior attempts to incorporate link dynamics in the modeling of routing and clustering schemes [25], [4] and [26], we believe that this new framework will find widespread use by researchers interested in the analytical modeling and optimization of MAC and routing protocols in MANETs. The advantage of our framework is that it accurately describes link, path, and topology dynamics as a function of node mobility. We then resort to simulation studies to pinpoint how mobility changes the performance of the underlying MAC layer. Finally, we strive to give a deeper insight on protocol operations by combining the analytical model and the simulated characterization of MAC layer performance.