[Paper Review] Gossip Based Ad-Hoc Routing

Zygmunt J. Haas, Joseph Y. Halpern+1 more|ArXiv.org|Sep 5, 2002

Mobile Ad Hoc Networks17 references4 citations

TL;DR

This paper proposes a gossip-based ad hoc routing protocol that reduces control message overhead by having nodes probabilistically forward routing messages instead of flooding. Using gossiping with probabilities between 0.6 and 0.8, the protocol achieves up to 35% fewer messages than flooding while ensuring most nodes receive messages in almost all executions, significantly improving performance in AODV and ZRP simulations.

ABSTRACT

Many ad hoc routing protocols are based on some variant of flooding. Despite various optimizations, many routing messages are propagated unnecessarily. We propose a gossiping-based approach, where each node forwards a message with some probability, to reduce the overhead of the routing protocols. Gossiping exhibits bimodal behavior in sufficiently large networks: in some executions, the gossip dies out quickly and hardly any node gets the message; in the remaining executions, a substantial fraction of the nodes gets the message. The fraction of executions in which most nodes get the message depends on the gossiping probability and the topology of the network. In the networks we have considered, using gossiping probability between 0.6 and 0.8 suffices to ensure that almost every node gets the message in almost every execution. For large networks, this simple gossiping protocol uses up to 35% fewer messages than flooding, with improved performance. Gossiping can also be combined with various optimizations of flooding to yield further benefits. Simulations show that adding gossiping to AODV results in significant performance improvement, even in networks as small as 150 nodes. We expect that the improvement should be even more significant in larger networks.

Motivation & Objective

• To reduce unnecessary routing message propagation in ad hoc networks, which suffer from high control overhead due to flooding-based protocols.
• To investigate whether probabilistic message forwarding (gossiping) can reduce routing overhead while maintaining high delivery rates.
• To evaluate the performance of gossiping in combination with existing protocols like AODV and ZRP, especially in large-scale networks.
• To understand the bimodal behavior of gossiping in large networks and identify optimal gossiping probabilities for reliable delivery.
• To explore the integration of adaptive gossiping probabilities based on route request success/failure feedback.

Proposed method

• Each node forwards a received routing message with a fixed probability p, replacing deterministic flooding with probabilistic forwarding.
• The protocol leverages percolation theory to analyze the bimodal behavior of gossiping: either the message dies out quickly or spreads to a large fraction of nodes.
• The gossiping probability p is tuned empirically between 0.6 and 0.8 to maximize delivery while minimizing message overhead.
• The approach is integrated into AODV and ZRP by replacing full flooding of route requests with probabilistic forwarding to peripheral nodes.
• Adaptive gossiping is explored, where the forwarding probability is adjusted based on route request success/failure and propagated via route request packets.
• Simulations are conducted on networks of varying sizes (including 150 nodes) to evaluate message count, end-to-end latency, and throughput.

Experimental results

Research questions

• RQ1 Can probabilistic message forwarding (gossiping) reduce routing control overhead in ad hoc networks without sacrificing delivery reliability?
• RQ2 What is the optimal gossiping probability p that balances message reduction and delivery coverage in large ad hoc networks?
• RQ3 How does gossiping compare to flooding in terms of end-to-end latency and throughput in AODV and ZRP?
• RQ4 Does the bimodal behavior of gossiping—where messages either die out or spread widely—enable predictable and scalable routing in large networks?
• RQ5 Can adaptive gossiping, where the forwarding probability is adjusted based on network feedback, further improve performance?

Key findings

• Using a gossiping probability between 0.6 and 0.8 ensures that in almost all executions, nearly all nodes receive the message, while reducing message overhead by up to 35% compared to flooding.
• In simulations, adding gossiping to AODV results in significant performance improvements in message count, end-to-end latency, and throughput, even in networks as small as 150 nodes.
• The performance improvement of AODV with gossiping is expected to increase in larger networks, where flooding overhead grows disproportionately.
• Gossiping can be effectively combined with existing optimizations, such as the cluster-based scheme in ZRP, to further reduce control traffic.
• The bimodal behavior of gossiping, predicted by percolation theory, is observed in simulations: either the message dies out quickly or spreads to a large fraction of nodes.
• Adaptive gossiping, where the forwarding probability is adjusted based on route request success, shows promise in improving performance, though optimal adaptation strategies require further study.