mac and routing interactions in mobile ad hoc networks


Overview - Description




The topology of a Mobile Ad hoc Network (MANET) is inherently unstable, due to mobility of the nodes that make up the network. However, several works have indicated that this issue is actually complicated by the fact that the IEEE 802.11 Medium Access Control (MAC) layer is poorly suited for use in these networks. Contending nodes often interfere with each other, causing increased instability to the network topology. These frequent changes in topology wreak havoc upon the routing protocol.

A large number of works exist relating to routing for the MANET environment. Although each protocol has a slightly different focus within the overall problem, the dynamic topology is necessarily always central. Every protocol must find a way to deal with managing the large number of changes in the network links, and therefore the changes in the network paths. While working protocols have been developed, they all suffer from one or more significant weaknesses, particularly delay or overhead expense.

In this project, we are working to draw benefit from the interactions of the MAC and routing in a MANET. Rather than having the MAC negatively impact on the routing protocol, we intend to exploit some of this closeness. We are developing an improved MAC layer for ad hoc networks. The added facilities within the new MAC can then be utilized in order to streamline the routing process.




A MANET is a collection of mobile devices (nodes) equipped with transmitters and receivers for wireless communication. These nodes can communicate with each other, however, they are limited by their transmitting and receiving capabilities (and network characteristics). Therefore, they cannot directly reach all of the nodes in the network as most of the nodes are outside of direct range.

In order to overcome this, the network operates in a multihop fashion. Nodes route traffic for each other. Therefore, in a connected ad hoc network, a packet can travel from any source to its destination either directly, or through some set of intermediate packet forwarding nodes. However, the mobility of the nodes can cause the network to change rapidly, resulting in a high volatility of the paths between any two nodes. This is the central problem in MANET routing.

Despite the large volume of work that has been published on routing protocols for ad hoc networks, no outstanding solution has been found. These protocols can basically be divided into three classifications: proactive (table-driven), on-demand (reactive), or hybrid (a combination of the previous two). Proactive protocols suffer from heavy overhead demands incurred by sending frequent updates in order to maintain consistent routing tables under the highly dynamic conditions. On-demand protocols often require a significant delay, as a path is discovered only when a source decides it needs to send to a particular destination. Hybrid protocols try to combine the features of each type of protocol, in order to minimize the drawbacks.

This hybrid category is the newest class of routing protocols for ad hoc networks. Likely the most recognized protocol of this type is the Zone Routing Protocol (ZRP). ZRP makes use of proactive routing in order to keep, maintain, and utilize the paths to nearby nodes (nodes within the local zone). Due to their proximity, the paths to these nodes can be kept relatively easily and inexpensively. For more distant nodes, source-initiated on-demand routing is used. Although this requires additional delay in order to discover the route, it eliminates wasted routing effort to these nodes.

Our work also falls into the hybrid category. It is dependent on several characteristics of an ad hoc network. First, the extreme volatility of paths within the network means that pre-building routes can be an expensive and wasted operation. Paths can break even before the process of on-demand route discovery is completed, or the usable lifespan may be very short. In addition, as longer paths include a greater number of links, their average lifetime also tends to be shorter. This means the added effort in building a longer path often needs to be performed more often.

Second, because all of the nodes in the network are capable of routing traffic, a number of different paths are available between source and destination pairs. Some of these paths may exist for a very short time, but some may be longer lived. Unfortunately, it is difficult to predict which will be the longest surviving paths at the time of route determination. However, being able to make use of these change-resistant paths may have great advantage over the more volatile paths.

Our approach has two key features. First, it keeps network-wide routing information, but only information that is less volatile than path information. It also does not require the information to remain precisely consistent. Plus, it is information that is often easily gathered, with minimal overhead. Second, route determination, based on this information, is left until the last possible moment, on a per-packet basis. This eliminates the possibility for paths to become stale, as well as allowing packets to follow the best-available path at the time they are being sent.

In order to allow this type of routing to be performed in an efficient manner, we have realized that we need some additional facilities at the MAC layer. We are attempting to alter the 802.11 MAC, adding these features while maintaining compatibility with the existing protocol. These additional capabilities basically permit the routing protocol to select an available next hop at the time of packet transmission.

We are currently working on simulating these ideas using the NS-2 simulator.