Implementation of Data Aggregation Method to Overcome Collusion Attacks in Wireless Sensor Networks

Abstract

Aggregation of data from various sensor nodes is typically done by basic strategies, for example, averaging or, more advanced, iterative filtering techniques. In any case, such aggregation strategies are profoundly defenseless against malignant assaults where the aggressor knows about every single detected esteem and has capacity to change a portion of the readings. In this work, we create and assess algorithms that dispense with or limit the impact of changed readings. The fundamental thought is to consider adjusted data as anomalies and discover algorithms that viably recognize modified data as exceptions and expel them. Once the anomalies have been evacuated, utilize some standard system to assess a genuine esteem. As the execution of low power processors significantly enhances, future aggregator nodes will be equipped for performing more advanced data aggregation algorithms, accordingly making WSN less helpless. Iterative filtering algorithms hold extraordinary guarantee for such a reason. Such algorithms all the while total data from numerous sources and give put stock in appraisal of these sources, more often than not in a type of comparing weight factors allocated to data gave by each source. In this paper we exhibit that few

 existing iterative filtering algorithms, while altogether more strong against agreement assaults than the basic averaging strategies, are all things considered susceptive to a novel advanced conspiracy assault we present. To address this security issue, limit esteem is expelled before registering an expected flag from the data focuses revealed by the sensors. In this manner, the proposed data aggregation algorithms works in two stages: expulsion of anomalies and algorithms of an expected genuine incentive from the rest of the sensor data. Broad assessments of the proposed algorithms demonstrate that they essentially beat every single existing strategy.