Modelling the Minimum Latency Aggregation Scheduling (MLAS) problem under traffic heterogeneity in clustered Wwireless sensor Networks.
Bawuna, Daniel Timothy
MetadataShow full item record
Data aggregation in WSNs, correlates and aggregates redundant or duplicate readings from sensors nodes before transmission to a designated destination. However, this inher- ently induces latency from transmission collisions causing large energy consumption in aggregation schedules. Scheduling collision-free network transmissions to minimize aggregation latency, known in academia as the Minimum Latency Aggregation Scheduling (MLAS) problem, has inspired a great deal of research. Past works on the MLAS problem however, have oftentimes only tackled aggregation scheduling of homogeneous data with no regard for heterogeneous data that requires prioritized data aggregation scheduling. Additionally, past efforts that tackled the MLAS problem, based solely on construction tree aggregation techniques which in combination ignored the energy-latency trade-offs and traffic heterogeneity. This study on the other hand, has considered energy-latency trade-offs for traffic heterogeneity while employing hierarchical clustering as an aggrega- tion technique. Our research derived an analytic model called the Bounded Aggregation- latency for Traffic Heterogeneity (BATH) that considers the two omitted research gaps. The BATH model dealt with prolonging network lifetime of a WSN by computing dynamic threshold bounds as a means of constraining the maximum latency of prioritized hetero- geneous traffic within optimal tolerances during aggregation scheduling. Formulations for these dynamic threshold bounds were derived based on the principle of conservation laws of queuing systems. This was applicable because, aggregation scheduling of het- erogeneous traffic required prioritized queues with aggregation service requirements that included SPTF, preemptive resume priority, non-preemptive priority and FCFS queues. Using parameterized MATLAB simulations, performance evaluation for the BATH model, was bench-marked on performance metrics of clustering as the underlying aggregation technique. Unlike past efforts that employed tree construction as an aggregation tech- nique with latency as their sole evaluation performance metric, clustering as an aggrega- tion technique allowed us other metrics necessary for latency-energy trade-offs. BATH model’s performance was evaluated against a generic aggregation scheduling model called Unbounded Aggregation-latency for Traffic Heterogeneity (UATH). This generic model was evaluated as a good baseline for comparison, reason being that UATH’s latency was not bounded. Numerical analysis of results for our research’s BATH model, showed a 1.2845 times increase in network lifetime compared to the UATH model. Prolonging network lifetime with such a margin similarly, showed improvement in the number of surviving nodes, clustering rounds and cluster head residual energy.