Mobile Ad-hoc network (MANET) is a collection of mobile nodes, which form a wireless network without the use of an existing infrastructure. MANET's are usually characterized by random mobility of nodes, nodes arbitrarily entering and leaving the network and variable transmission range of nodes. These characteristics make MANET links to be intermittent and the topology to be highly dynamic. In such situations routing protocols are desired to be on-demand and robust enough to cope up with the dynamic topology changes. Many on demand routing protocols have been suggested for this purpose like single path Dynamic source routing (DSR) and Ad-hoc on demand distance vector routing (AODV) and their multiple path variants like Ad-hoc on demand multiple distance vector routing. These protocols suggest different ways to avoid route-looping problems and create shortest path route. Locally adaptive and multi-metric Ad-hoc on demand distance vector routing (LAM-AOMDV) tries to use multiple metrics to discover routes and proposes new schemes to provide route maintenance. The objective of a routing protocol is to efficiently find routes and transfer as many data-packets possible with minimum routing overhead. In our thesis work propose two layers of improvement one to reduce the overhead caused unnecessary transmission of hello packets and second to pre-emptively discover routes in cases of a link breakage, which renders a path useless. In our schemes we try to stop the unnecessary hello transmissions by nodes, which are not participating in an active route or in route discovery. We also try to improve the packet delivery ratio by finding routes pre-emptively before the last route becomes unavailable to the source. We compare the individual layers of improvement with LAM-AOMDV using various metrics like packet delivery ratio, overhead caused due to hello messages and packets dropped due to no route availability at the routing layer.