چكيده به لاتين
The Internet of Things is a new technology that has extended rapidly and used in industry and urban life widely. Due to the increasing number of wireless objects, limited power of IoT devices and the effect of increasing energy consumption on environmental pollution, energy optimization has become the main challenge in the Internet of Things. One of the energy optimization methods is data aggregation, which means data volume reduction to reduce required energy for data process and transmission. There are three approaches in Data aggregation: centralized, cluster-based, and tree-based data aggregation. In centralized method, data aggregates in a central node that reduces accuracy and scalability. To solve these challenges, cluster-based method proposed in which nodes grouped into separate clusters and data for each cluster aggregated separately. This method has a long delay due to the aggregation of the entire data of a cluster in the cluster head. In addition, due to the importance of compatibility between cluster data type, it is necessary to form clusters before data aggregation that is proportional with data homogeneity. Therefore, clustering is one of the most important challenges of cluster-based aggregation. Due to the limitations of centralized and cluster-based approaches, today most of the research use tree-based approach that aggregates data by creating a hierarchical structure in the network.
In this study, tree-based data aggregation has used in order to energy optimization. Most of the presented methods based on tree-based data aggregation, only consider energy in their optimization goals. However, paying attention to the throughput optimization along with energy consumption will improve the quality of service besides improving quality of energy consumption. In addition, joint tree formation, routing and scheduling cause throughput has direct impact on forming aggregation tree. However, the numerated methods that address these two goals together do not support joint tree formation, routing, and scheduling. To solve this problem, the TART-Optimal model has proposed that simultaneously saves energy by forming a data aggregation tree, increases throughput by dividing time to slots and link scheduling. Due to the time complexity of this problem, we proposed the ant colony algorithm to solve it. In the proposed TART-ACO algorithm, the paths formation process will be interdependent due to the data aggregation in different paths. Therefore, in this algorithm, all ants work together to produce a unique scheduled tree. Evaluation results in the small-scale assessments show that the proposed TART-Optimal model and the TART-ACO algorithm perform better about 14% and 8%, respectively in obtaining scheduled aggregation trees compared to A Disjoint-Data Aggregation and Scheduling (D-DAS) model, according to the Mean Absolute Percentage Error Criteria. In addition, based on the the Mean Absoloute Error and the Root Mean Square error, the actual difference between the TART-ACO algorithm and the TART-Optimal model is about 0.1.
Keywords: Internet of Things, Energy Optimization, Ant Colony Algorithm, Tree-Based Data aggregation, link scheduling
