Doctoral student Yao Lu will defend his Ph.D. thesis on the topic:
Adaptive Data Aggregation based on Multi-Objective Ant Colony Optimization in Wireless Sensor Networks
Time: Friday (02.12.2016), 04:15pm-05:15pm
Location: Boulevard de Pérolles 90, Pérolles II, Building PER.21, A230
Admittance: free
Abstract In the data collection applications of Wireless Sensor Networks (WSNs), overhigh data redundancy and communication load are the most conventional challenges for the constrained sensor resources, especially power supply. In-network data aggregation is developed to address these problems through merging data messages during routing process. How to efficiently transport the designated sensor readings from source nodes to sink nodes by using data aggregation functions is the most important issue concerned in this thesis. From the spatial perspective, the explorations of the optimal routing structure for data aggregation can be defined as multi-objective combinatorial optimization problem. From the temporal perspective, the theoretical optimal waiting intervals of unslotted aggregation timer on each forwarding node should be analyzed as well. For the purpose of guiding the transmission direction of data messages and exploring the optimal routing structure, Ant Colony Optimization (ACO) is naturally embedded into the routing process. Genetic Algorithm (GA) is adopted to accelerate the convergence speed of structure exploration. In order to match the dynamic feature of routing scheme, an adaptive timing policy based on sliding windows is designed to control the time point of transmission and aggregation, the waiting interval of aggregation timer can be automatically adjusted according to the prediction of future messages. Furthermore, an extended version of aggregation approach working in a fully distributed manner is proposed to support the application scenarios with multiple sink nodes. Thanks to the broadcast nature of wireless communication, multicast transmissions are used to simultaneously exchange information with multiple targets. Finally, a periodical data collection system is implemented on the OMNeT++ simulator, and the simulation results validate the efficiency of our proposed approach.