On the Effects of Aggregation on Reliability in Sensor Networks

Abstract

Abstract-Data collected in a sensor network is transported hop-by-hop to a sink for further analysis. The quality of the analysis depends on the amount of data reaching the sink. Hence, data transport reliability influences the quality of the analysis. Data aggregation is a common method used in sensor networks to reduce the amount of messages transported. By aggregating, the data contained in several messages is fused into one single message. Therefore, data aggregation significantly influences the overall data transport reliability observed at the sink. This influence is analyzed and described analytically and by experiment within this paper. Furthermore it is shown how the influence of data aggregation on data transport reliability can be controlled for a particular class of data gathering application.

A Study on Data Aggregation and Reliability in Managing Wireless Sensor Networks

“…1A) the number of reports received at the sink is the same whether each report is sent individually, or all are aggregated into a single packet [15]. We now examine the situation where every node in the tree is a source (Fig.…”

“…For a WSN environment, data aggregation does not effect reliability in terms of the number of reports received at a root node from a source (the expected value) when compared to the non-data aggregation case [15]. But it does increase the variability.…”

Reliability Control for Aggregation in Wireless Sensor Networks

“…In contrast this paper relaxes the assumptions used and recalculates the necessary forwarding reliability to meet application targets as aggregation occurs in network. This offers a much more fine grained approach than [11] and also relaxes a number of assumptions used in that paper. An alternative to altering the forwarding reliability to suit the aggregation level would be to adjust the aggregation level to suit the inherent reliability of the links available.…”

“…Losing a message containing a single data reading has surely a different impact on the overall data reliability than losing a message containing the information of several sensor readings. This effect is described in [11]. Where aggregation occurs the average amount of data arriving at the sink, expressed by the expected value E(X) is unaltered compared to when aggregation is not used.…”

“…Worst case dimensioning was examined in [11] where it was found that, although the methods used were successful from a dimensioning point of view, significant overshooting of application defined targets occurred due to the use of worst case assumptions. In contrast this paper relaxes the assumptions used and recalculates the necessary forwarding reliability to meet application targets as aggregation occurs in network.…”