A routing-aware middleware-level mechanism for tuple-based pervasive systems

Existing tuple-based middleware-level mechanisms for pervasive distributed systems often use broadcasting for resource discovery. However, broadcasting increases the required query processing time for finding resources in the peers, leading to increased response time of pervasive systems. This paper presents a new routing-aware mechanism for the same purpose to attain better response time that is for the first time based on social peer to peer (S2P) networks that imitates human behaviors in social interactions. Instead of broadcasting a request tuple for a resource to all peers in a pervasive system, the request tuple is only passed to a single peer who has had the resource previously. This is achieved by introducing a simple data structure inside every peer in the system that contains information on which peers possessed which resources based on the findings of the same peer in its previous queries for resources. This information is in fact cached and used in the routing request tuples for resource discovery. Each peer caches and uses such information by itself for the purpose of routing quite independent of other peers to save communicating with other peers for routing its request tuples. We implemented the proposed mechanism in Java and ran it on a prototyped pervasive system under LIME environment in a MANET. The results of runs of our mechanism demonstrated an approximately 53% improvement in the query processing response time of the constructed pervasive system compared to the response time of the same system running the default routing mechanism in LIME.