We've Moved!
Visit SDSU’s new digital collections website at https://digitalcollections.sdsu.edu
Description
Due to the tremendous surge in IoT (internet of things) devices, traditional network protocols cannot accommodate high-throughput flows due to the limited channel capacity, which leads to congestion, thereby, resulting in buffer overflow and packet expiry due to high queuing delay. This phenomenon ultimately results in considerable degradation of flow throughput. Therefore, novel network protocols are needed in order to faciliate applications with high quality of service (QoS) requirements. A potential solution to accommodate flows at a high data rate is to use multiple paths for load balancing. In this thesis, we introduce two multipath node disjoint routing mechanisms in the OLSR (optimized link state routing) protocol in the network simulation tool NS-3 and analyzed their performance for different network topologies and at different data rates. Our proposed scheme has achieved a significantly higher flow throughput and a lower end-to-end delay compared with the existing single-path OLSR scheme. The main contributions of this thesis are as follows: (i) A multipath Dijkstra algorithm is implemented to identify multiple node disjoint paths at a node using the information stored in a routing table. (ii) A source routing mechanism is implemented, which allows an intermediate node to select between a globally feasible route (found at the source node) and a locally best next-hop node (found at that intermediate node). (iii) A heuristic mechanism is proposed for proactive buffer management, which proactively drops packets that are likely to be dropped due to TTL (time-to-live) expiry before reaching the destination node. This mechanism reduces network congestion and resource wastage.
