Distributed Split-Path Routing Strategy for Multi-hop Mesh Networks Paulo Alexandre Regis, Shamik Sengupta Department of Computer Science and Engineering University of Nevada, Reno, USA, 89557 pregis@nevada.unr.edu, ssengupta@unr.edu Abstract—The concept of mesh networks brings many chal- lenges to the research community when designing such system. In this paper, we propose a new routing model to account for the link utilization between nodes, as well as the remaining energy left in each node. The framework, called Distributed Split-path, is implemented on top of the Optimized Link-State Routing (OLSR) protocol. We demonstrate the improvement our proposed scheme offers to the longevity of the network while maintaining similar performance as the original OLSR protocol. We devise a performance indicator to show in which cases our approach benefits the network system designer, depending on the priorities of the network application. Simulation results show that by allowing a trade-off in the performance, the new model outperforms current protocol. Keywords—Wireless mesh network, routing, metric, energy efficient, multi-path I. I NTRODUCTION Due to the rapid evolution of handheld electronic devices, a massive surge in wireless communication demand was seen in the last decade. Together with the development of small unmanned aerial vehicles (UAVs), this advancement brought a new age of mobile ad-hoc networks (MANET) applications. The new UAV technologies, coupled with developments in unmanned ground vehicles (UGVs) added a new dimension to service provision of wireless networking systems. The new trend of 3D wireless mesh networks enables mission-centric operations like military tactical network and first responders in disaster areas network in situations where infrastructure is not available. Throughput hungry applications, such as real-time video streaming and large files transfer, push forward the develop- ment of higher capacity wireless networks. One alternative is to distribute the load in the network among all of its nodes. This way, if well designed, the network reduces the traffic bottleneck while maintaining the performance at an acceptable level. When supporting a specific mission, the priority of the network is to provide uninterrupted communication, as well as offer enough capacity to execute the task. Mesh networks can provide a redundant connection by forming multiple paths between two nodes, this way if one path is disrupted, communication can be rerouted through a different set of nodes. Simple examples are border control and surveillance networks. Sensors are deployed to collect certain types of data and send back to a control base, where the data can be analyzed. These sensors are typically small and only perform a single specific task (measure, record, film, etc.). They can be deployed in areas with difficult access, which makes main- tenance a challenge. Thus, creating mechanisms to maximize the longevity of the devices is very important. Devices capable of forwarding packets can autonomously create a multi-hop network to provide wireless coverage to tactical or disaster teams over a wide area by relaying the data through multiple Figure 1: Example application scenario: arrows represent a possible path of communication between end-nodes. The different colors (red and black) distinguishes different com- munication channels nodes and frequencies. We see an example of such an appli- cation in Figure 1. The UAV network provides connectivity to the soldiers when there is no infrastructure available. It enables the team to communicate among themselves or send information to a remote location, such as the headquarters, through a gateway in the network. Having more than a single path of communication makes the system more robust in case of a node malfunction. The multi-hop nature of the network poses some serious challenges for resource (time, bandwidth, power) allocation among the devices. Ideally, the network should not have any bottleneck to avoid interruption of the service, and nodes should be able to function for as long as possible before running out of energy. The longevity problem becomes very complicated if no centralized controller defines the positioning of nodes, and the paths the end-to-end communication should traverse, based on the information of the entire network. However, having a complete information of the network might not even be feasible, due to the dynamic and autonomous characteristics of a mesh network. In a network with energy constrained devices, increasing the throughput without con- cerning about the individual power consumption of the nodes can interrupt the connectivity of the entire system if there is a bottleneck (i.e. most of the communications have to pass through a small set of nodes). In this paper, we expand our previous work where we designed a strategy to pair traffic flows and paths in an ad- hoc network [1]. We develop a new routing mechanism that encompasses information from other layers of the network stack. The new approach, called Distributed Split-path, chooses the next hop for a packet based on the previous route requests, combined with the local energy information and transmission characteristics of the link. The traffic should be split into multiple paths to improve the performance regarding energy