A cross layer routing metric with wireless cooperative protocols

Cooperative link layer protocols are typically used in sing le hop networks. In such protocols, a special node called the relay node helps deliver frames from a source to a destination. The performance benefits of cooperation at link layer can be stre amlined into multi-hop networks as well. In multi-hop networks, a fr me is sent from an original source to the final destination th rough a series ofintermediate nodes. The paper extends the expected transmission time metric — p roposed for multi-hop wireless ad hoc networks — to the context of cooperative IEEE 802.11 link lay er protocol. The designed metric is called cooperative expe cted transmission time (CETT). CETT carefully accounts for the h igher probability of successful frame transmission and the refore the reduction in expected transmission time brought about by th e relay node in the cooperative protocol. CETT jointly optim izes both the route computation and the selection of the cooperative r elay at the link layer. Route optimization helps jointly cho ose the best set of intermediate nodes and cooperation optimizatio n helps choose the best relay node for each link in the multi-h op. As a result, CETT helps distinguish the case wherein it is bette r to use a node as a relay as compared to using it as an intermedi ate node. For comparison, the case where cooperation is applied aft r route computation is also presented. Minimizing the e xpected transmission time may result in more efficient link utilizat ion and increased overall end-to-end network throughput. I t is also shown that joint optimization of route and relay selection i s better than finding routes and then applying cooperation. I. I NTRODUCTION Wireless channels are susceptible to fluctuations in terms o f path loss, fading, etc. Wireless link layer protocols prov ide automatic retransmission request (ARQ) mechanisms to coun ter frame corruptions due to both frame errors and collision s. For example, in IEEE 802.11 medium access control (MAC) protoco l, a transmitter can attempt to transmit a given frame multip le times, up to a predetermined maximum value, before moving to the next frame awaiting transmission. ARQ mechanism may be enhanced using cooperative nodes, as proposed by a number of auth rs [1], [2], [3]. In link layer cooperative ARQ protoco ls, a third node, namely the relay, helps deliver frames from source to destination. While the relay’s intervention is regulated by specific rules — which vary from protocol to protocol — its coo peration can bring perceivable performance advantages in single hop networks [1], [2], [3]. Designing routing protocols for multi-hop wireless networ ks constitutes yet another challenge in terms of overhead in curred in establishing and maintaining end-to-end routes along mult iple intermediate nodes. Several routing protocols have been proposed in recent years, e.g., [4], [5]. These routing protocols try to minimize the hop count — i.e., the number of intermediate n odes — in reaching the destination. It is well understood that the hop count metric may not be the best choice for wireless netwo rks as multiple transmission rates are now available and differ ent error probabilities for the corresponding rates may be e xpected. Some work has been done in this regard [6], [7]. One such propo sed work — namely the expected transmission time (ETT) — attempts to minimize the expected air time that is consumed in successfully delivering a frame from the source to the end destination. Another metric — expected transmission ti me (ETX) — attempts to minimize the number of transmissions required to deliver a packet to the end destination. It is not ed that ETT is bandwidth adjusted ETX [7]. The relay node can help lower this expected transmission tim e of a frame. Network layer may or may not be aware of the link layer cooperative protocol. Both these cases have been explored in this paper. In the case where network layer knows the existence of relays, the links in a route can be stretched as t he relay node would compensate frame losses. On the other han d, when the network layer is unaware of the relays, route comput ation is done independent of relays and therefore converges to the same path as the non-cooperative case. However, link lay er finds a suitable relay for each link. When the network layer is aware of the underlying cooperativ e link layer protocol, there is no reason to exclude that cooperative protocols at the link layer and routing protoco ls may coexist in the same network. There is no evidence that the two combined network functionalities can yield any mean ingful performance advantage. In fact, one could argue that in multi-hop networking a potential relay node may better func tion as intermediate node along the route. In this paper the authors present a study to clarify this poin t, a d possibly reach a conclusion about the usefulness of cooperative link layer protocols in multi-hop networks. Th eorems are presented to demonstrate that not all cooperativ e link This research is supported in part by NSF grants No. ECS-0225 528 and CNS-0435429