Augur: a delay aware forwarding protocol for delay-tolerant networks

Abstract

Real life applications of Delay-tolerant Networks (DTNs) are evolving rapidly in many medical, environmental and engineering fields. DTN supports networking where the existence of contemporaneous communication links between the nodes is not guaranteed.In fact, DTN nodes are usually deployed in extreme terrestrial or spatial environments that lack continuous network connectivity. The nodes can range from sophisticated devices with abundant resources such as those used in interplanetary networks to small devices that are very limited in resources such as those operating in wireless sensor networks.Moreover, the mobility of the nodes and other disruption factors that may be present in the field cause frequent disconnections and nodes isolations leading to an intermittent connectivity.An efficient DTN routing protocol should have two main characteristics: high delivery probability and low delivery delay. Many routing algorithms have been proposed in various DTN applications. The majority of these protocols optimize delivery probability whereas very few of these algorithms address low delivery delay. Specifically, no previous work has produced a delay aware protocol to route data in DTN scenarios using historical spatiotemporal data of mobile nodes.Spatio-temporal information and encounter statistics provide useful measures to understand a node's mobility. The time dependent behaviour of a mobile node and its periodic reappearances at specific locations around similar times for similar durations can predict future presence of that node. This characteristic can assist the performance of a routing protocol by estimating the time, location and duration of possible upcoming transfer opportunities.This thesis addresses the delay issue in DTN by studying the effect of including the spatial and temporal dimensions of mobility in the decision metric of DTN protocols. "Augur" a new delay aware routing protocol for DTNs is introduced. In particular, Augur is targeted to optimize and minimize message delivery delay based on historical spatiotemporal behaviour of the participating nodes.Two versions of Augur will be presented in this work namely Augur Temporal and Augur SpatioTemporal. The two versions differ in the amount of historical information which is used regarding the movement characteristics of the mobile nodes. Augur Temporal uses only the temporal dimension to build its decision metric. This protocol is targeted to nodes having highly repetitive and periodic movements every day. The SpatioTemporal version makes use of both space and time dimensions to suit high and low periodicity scenarios at the expense of more processing and storage.This work investigates the performance of the proposed Augur algorithms in an application related to mobile ad hoc networks. It compares their performance to the state of the art DTN protocols using the same set of parameters; e.g., number of nodes, load, buffer size and movement model. The implemented scenario considers a Helsinki city map in which bus nodes move along predefin...