Asymptotics of query strategies over a sensor network

Abstract

We consider the problem of a user querying for information over a sensor network, where the user does not have prior knowledge of the location of the information. We consider three information query strategies: (i) a Source-only search, where the source (user) tries to locate the destination by initiating query which propagates as a continuous time random walk (Brownian motion); (ii) a Source and Receiver Driven "Sticky" Search, where both the source and the destination send a query or an advertisement, and these leave a "sticky" trail to aid in locating the destination; and (iii) where the destination information is spatially cached (i.e., repeated over space), and the source tries to locate any one of the caches. After a random interval of time with average t, if the information is not located, the query times-out, and the search is unsuccessful.For a source-only search, we show that the probability that a query is unsuccessful decays as (log(t)) −1 . When both the source and the destination send queries or advertisements, we show that the probability that a query is unsuccessful decays as t −5/8 . Further, faster polynomial decay rates can be achieved by using a finite number of queries or advertisements. Finally, when a spatially periodic cache is employed, we show that the probability that a query is unsuccessful decays no faster than t −1 . Thus, we can match the decay rates of the source and the destination driven search with that of a spatial caching strategy by using an appropriate number of queries.The sticky search as well as caching utilize memory that is spatially distributed over the network. We show that spreading the memory over space leads to a decrease in memory requirement, while maintaining a polynomial decay in the query failure probability. In particular, we show that the memory requirement for spatial caching is larger (in an order sense) than that for sticky searches. This indicates that the appropriate strategy for querying over large sensor networks with little infrastructure support would be to use multiple queries and advertisements using the sticky search strategy. *

Broadcasting in Sensor Networks: The Role of Local Information

“…They have shown that querying delay, transmission overheads can be reduced by spreading routing information through the network. This phenomenon has been quantitatively studied in [17].…”

Energy Efficient Data Acquistion in Wireless Sensor Network

“…In sensor networks, data caching has one more important benefit that is saving communication energy cost. Many existing research works focused on strategies of replicating frequently accessed sensed readings in some sensor nodes closer to base stations Ganesan et al (2003); Liu et al (2004); Ratnasamy et al (2002); Sadagopan et al (2003); Shakkottai (2004); Zhang et al (2007). In presence of multiple base stations, a research problem of finding sensor nodes for caching sensed readings is formulated as determining a Steiner tree in a sensor network Prabh & Abdelzaher (2005).…”

Sparse Data Aggregation in Sensor Networks

“…Different from a pure 'pull' or pure 'push' approach (both of which involve expensive flooding operations), a double ruling scheme adopts a symmetric information discovery principle with both the source and sink actively searching for each other, as first suggested in rumor-routing [1]. The notion of trails has appeared in a number of other papers as well, including in trajectory forwarding [14], in the asymptotics of query strategies [16], and the combs and needles work [10]. A number of double ruling schemes that guarantee distance sensitivity have been devised in the literature [15].…”