Article Preview
TopIntroduction
Ad hoc wireless networks are multi-hop networks without fixed infrastructure. Each node in such a network can act as a router to forward data for other pairs of nodes which are outside of each other’s transmission range. Congestion is an important challenge faced by such networks.
Many studies have investigated congestion and AODV modification to improve network performance in ad hoc (wireless) networks. E.g., Senthil and Sankaranarayanan (2011) propose EDAODV (early detection congestion and control routing protocol), which computes congestion status from queue status, defining both a minimum (25% of buffer size) and a maximum threshold (3 times minimum threshold) to classify nodes according to congestion level. Nodes in the current route detect the congestion status and periodically broadcast CSPs (congestion status packets) to announce their congestion status. If the uncongested neighbours receive a CSP from a node, EDAODV initiates a bi-directional route discovery process to find an optimally uncongested path between the two neighbours. Note that each node in EDAODV monitors its own congestion status and needs to notify its neighbours of congestion via the CSP broadcast.
Also based on AODV, Cuong et al. (2016) suggest a cross-layer routing protocol, AODV-DM (Ad hoc On Demand Distance Vector-Delay Metric). They observe that the average link delay of a congested area should exceed that of an uncongested area and estimate the link delay at the MAC layer based on the service time. AODV-DM then applies a cross-layer approach to use this delay value as the routing metric at the network layer to find a route that avoids the congested area. However, even though link delay can be used to distinguish whether an area is congested or not, delay may vary even for nodes using the same route for data transmission in ad hoc networks since wireless nodes do not have a universal time standard.
Jyoti et al. (2013) modify the route maintenance process of AODV and propose a novel method to repair a link locally. Instead of periodically broadcasting Hello messages to check node connectivity, they maintain the connectivity of nodes via the 802.11b protocol. They try to repair broken links by finding the node beyond the next node, which is provided via RREP messages. But this local repair method could be considered to only work for ad hoc networks with 802.11b protocol. Also, the node(s) which is/are used to repair broken links need(s) extra protocol overhead.
Kaur and Pandey (2015) also modify AODV and introduce M-AODV to find an optimal path to decrease network congestion. Rather than considering the shortest route as AODC does, they use delay and energy consumption of nodes as well as the lifetime of the network as parameters in the selection of a route that avoids congested areas and reduces congestion as far as possible. Hence, M-AODV also uses the congested nodes themselves to manage congestion with a pretty complicated mechanism to determine congestion. Again based on AODV, Dipika et al. (2018) present a new route selection algorithm that leverages ACO (Ant Colony Optimization) to select an optimal route for ad hoc networks. For the ant colony algorithm, they assume that the best data delivery route is that with the highest pheromone value. The pheromone value is resulted from the end-to-end reliability of the route, the congestion among nodes, the number of hops and the residual energy of the nodes along the route. Dipika et al. also use buffer occupancy as an indicator of congestion and let the congested nodes themselves take care of congestion resolution.