|
Epidemic protocols belong to a class of routing paradigm that have wide ranging applications in Delay Tolerant Networks (DTNs) due to their simplicity, low delays, and little to no reliance on special nodes. To this end, a comprehensive study of their performance will serve as an important guide to future protocol designers. Unfortunately, to date, there is no work that studies epidemic routing protocols using a common framework that evaluates their performance objectively using the same mobility model and parameters. To this end, we study four categories of epidemic routing protocols. Namely, P-Q epidemic, epidemic with Time-To-Live (TTL), epidemic with Encounter Count (EC) and epidemic with immunity table. Our results show that the probability of transmissions as used in P-Q epidemic may increase delay and decrease delivery ratio. Apart from that, an incorrect TTL value leads to premature discarding of bundles, and thereby, has a non negligible impact on delivery ratio. Epidemic with EC suffers from high buffer occupancy levels and long delivery delays. In addition, epidemic with immunity suffers from high overheads. Henceforth, we propose three enhancements: dynamic TTL, EC+TTL and cumulative immunity to address the aforementioned limitations. Our results show that dynamic TTL improves delivery ratio by more than 20\%, EC+TTL reduces buffer occupancy level by 40\%, and improve delivery ratio by at least 40\% at high loads. Cumulative immunity reduces nodes buffer occupancy level by at least 15\% whilst incuring an order of magnitude less signaling overheads.
|