Pyrih Y. Models and algorithms for data flows routing in self-organized networks.

The thesis is devoted to solving the actual scientific task of the data flows routing methods enhancement in self-organized networks by improvement of the data transmission latency in conditions of the dynamic traffic intensity and noise fluctuations in modern wireless sensor networks in order to satisfy user requirements in real-time applications. The principles and features of the wireless self-organizing networks design have been analysed to determine the key tasks, which should be solved to improve the balance and fairness of radio resource utilization such as improvement of the routing efficiency and spectrum sharing in licensed bands without jeopardizing the user experience. In order to solve the given tasks this thesis propose new complex routing method based on the global extremum determination for the function of sorted random search by adding/removing of the nodes. Proposed method allows creating and arranging of the set of routes by using the QoS maximization and taking into account the network heterogeneity. The nodes localization method has been improved based on the Thurston's algorithm using the Voronoi power diagram with correct adjacent connections, which allows to increase the precision of virtual coordinates by using canonical unit Poincare disk to improve the quality of routing in Euclidian space. The clustering method of sensor nodes has been improved by the determination of centroid taking into account the wireless channel propagation model to reduce the time for route searching between the arbitrary pair of nodes. The model of clusters overlapping in self-organized wireless networks has been further developed by using the logical nodes addresses reassignment within the given cluster to ensure the target QoS level in case of hardware or software failures. The simulations have been conducted based on the developed analytical and empirical models in conditions of noise fluctuations in wireless channels to prove the efficiency of the proposed solutions. Simulation results show that by using the proposed complex routing method we can ensure the high reliability of the wireless sensor network and maintain the required QoS level for end users. Simulations on the efficiency of proposed clustering method show that bit error rate has been reduced by 7% comparing to existing solutions. Simulations of the transmission time between source node and destination node using the shortest path selection by the proposed complex routing method show that the transmission latency has been reduced two times comparing to conventional routing methods for the same configuration of the wireless sensor network.