Data Aggregation Algorithms with Multiple Sensors in Clustered-Based WSN/IoT

Abstract

In wireless sensor networks (WSN), data aggregation algorithms are used to extend the network lifetime. The size of data packet transmitted from the cluster head (CH) to the base station (BS) seriously affected energy consumption in a CH. In this paper, three types of data aggregation algorithms are evaluated. These techniques are coding schemes based relative difference (CS-RD); adaptive data aggregation method (ADAM); coding schemes based on the factor of precision (CS-PF). The performances of the algorithms are compared based on 15 different scenarios. The algorithms are applied separately with the following parameters: (1) Mean; (2) Median; (3) Mode; (4) Geometric mean; (5) Harmonic mean. Experimental efforts are taken on each scenario separately for the multiple sensors recording the temperature, humidity, and light. The performance metrics studied are energy consumption, average of absolute error and data compression ratio. The simulation results showed that the best performance is shown by the CS-RD algorithm. The ADAM produces an intermediate performance for all sensors. Overall, it can be said that the accuracy of CD-PF is better in comparison with other algorithms. Nevertheless, it displays worst performance in energy consumption and data compression ratio for all scenarios. From the results, it was noted that the selection mechanism suitable to determine the central point effect is based on the performance of the three aggregation algorithms. For temperature (T) and humidity (H) sensors the best performance in terms of energy consumption is the CS-RD, which is less than 800 uJ and compression ratio of more than 90%. ADAM and CS-FP algorithm with Mean /Gmean /Hmean methods showed better performance with energy consumption of less than 1000 uJ and compression ratio of 91%. Mode method negatively affected the performance of all algorithms, with CS-RD energy consumption reaching 2500 uJ. Finally, for light sensor, the CS-RD shows best performance with all central point methods, where the energy consumption goes below 1400 uJ. The CS-PF and ADAM with mode showed the highest energy consumption higher than 4200 uJ and 2400 uJ, respectively.