<XML><RECORDS><RECORD><REFERENCE_TYPE>10</REFERENCE_TYPE><REFNUM>9130</REFNUM><AUTHORS><AUTHOR>Koliousis,A.</AUTHOR><AUTHOR>Sventek,J.</AUTHOR></AUTHORS><YEAR>2007</YEAR><TITLE>Proactive vs reactive routing for wireless sensor networks</TITLE><PLACE_PUBLISHED>DCS Technical Report Series</PLACE_PUBLISHED><PUBLISHER>Dept of Computing Science, University of Glasgow</PUBLISHER><ISBN>TR-2007-293</ISBN><LABEL>Koliousis:2007:9130</LABEL><KEYWORDS><KEYWORD>proactive routing</KEYWORD></KEYWORDS<ABSTRACT>Sensor network routing protocols must ensure the stability of the network infrastructure under varying network dynamics. Recovery from changes or failures is necessary to guarantee the availability of collection or dissemination paths. Routing protocols may react to path requests or proactively maintain a connected graph. Although functionally equivalent, proactive and reactive protocols are associated with different costs, in terms of resource overhead (e.g. energy or bandwidth) and non-functional guarantees (e.g. end-to-end delay, or time to repair). The protocol of choice must satisfy the system objectives, yet it should not require excessive resources. We study the behavior of proactive and reactive routing schemes, using prototype TinyOS implementations of the OSPF and AODV protocols, respectively. Simulation results show that there exist regions in the time-to-repair x overhead space where reactive protocols are preferred over proactive and vice versa; the proactive preference region grows as the number of simultaneous flows increases.</ABSTRACT></RECORD></RECORDS></XML>