Good Route Hunting : Improving Routing Efficiency in Large-scale Tactical OLSR Networks

Sammanfattning

The ability to evaluate tactical Mobile Ad hoc Networks (MANETs) and their protocols through simulation is an important instrument for the military sector as it saves significant time and resources compared to live testing. Optimized Link State Routing (OLSR) is one such protocol that has long been of interest for tactical use, with some reservations regarding scalability. The Swedish Defence Research Agency (FOI) uses the simulation tool Aquarius to evaluate these types of radio communication networks, but in order to conduct large-scale OLSR simulations, a more efficient routing algorithm needs to be implemented. This study has examined the currently implemented OLSR routing algorithm and improved upon the inefficiencies found, primarily relating to the data structures used. The performance of the original algorithm, together with a few modified versions, was experimentally evaluated using different scenarios, some in realistic, tactical settings. The most optimized version(s) outperformed the original algorithm across all scenarios and network conditions, demonstrating superior routing speed and thus scalability. An approach to dynamically apply link changes was also tested to avoid reconstructing the entire network graph before each routing calculation as per the OLSR standard, shedding light on the challenges associated with converting and storing data from the information repositories over sequential routing calculations. Experimental evaluation yielded additional improvements with regards to routing time, the increased memory usage (presumably) did however have consequences for the overall execution time of the simulations, resulting in unpredictable behaviour, requiring further investigation.