TL;DR
This paper comprehensively evaluates and compares six popular wireless routing protocols (three reactive and three proactive) under various mobility scenarios. It reveals that reactive protocols generally outperform proactive ones in terms of throughput, especially under high mobility, but often at the cost of increased delay.
What Problem Does This Paper Address?
In mobile ad-hoc networks (MANETs), nodes are constantly moving, leading to dynamic network topologies. Routing protocols are essential for establishing and maintaining communication paths in these networks. However, the effectiveness of a routing protocol can be significantly impacted by node mobility and speed. This paper addresses the need for a clear understanding of how different routing protocols, specifically reactive (on-demand) and proactive (table-driven) types, perform under varying mobility conditions. The authors aim to provide a comparative analysis to help researchers and engineers choose the most suitable protocol for their specific application needs, considering trade-offs between throughput, delay, and routing overhead.
Methods
The authors employed extensive simulations using the Network Simulator 2 (NS-2) to evaluate the performance of six routing protocols:
- Reactive Protocols: Ad hoc On-Demand Distance Vector (AODV), Dynamic Source Routing (DSR), and DYnamic MANET On-demand (DYMO).
- Proactive Protocols: Destination Sequenced Distance Vector (DSDV), Fish-eye State Routing (FSR), and Optimized Link State Routing (OLSR).
They utilized four established mobility models: Fast Car Model (FCM), Slow Car Model (SCM), Human Running Model (HRM), and Human Walking Model (HWM), which represent different speed ranges. Simulations were run for various pause times (from 0s to 900s) to represent periods of node rest. The performance was evaluated based on three key metrics:
- Throughput: The amount of data successfully transferred.
- End-to-End Delay (E2ED): The time taken for a packet to travel from source to destination.
- Normalized Routing Load (NRL): The number of routing packets transmitted per successfully delivered data packet.
Datasets
Not discussed in the paper. The evaluation was based on extensive simulations using NS-2.
Key Insights and Ideas
- Reactive vs. Proactive Trade-off: Reactive protocols generally offer better throughput, especially under high mobility, because they only establish routes when needed, avoiding the overhead of maintaining stale routes. However, this often comes at the cost of higher end-to-end delay.
- DSR's Resilience: DSR demonstrated strong performance, particularly in terms of throughput, due to its route cache mechanism that stores multiple routes and its promiscuous listening mode, which helps it adapt to frequent link breakages.
- AODV's Balance: AODV offers a good balance, handling various mobility rates and traffic levels, though it can incur more routing load than DSR due to its HELLO messages and local link repair mechanisms.
- DYMO's Underperformance: DYMO showed the lowest throughput among reactive protocols, likely due to its less sophisticated route discovery and maintenance mechanisms compared to AODV and DSR.
- DSDV's Strength in Low Mobility: DSDV, a proactive protocol, achieved high throughput in scenarios with low mobility and static conditions due to its stable routing tables.
- OLSR's Low Delay: OLSR consistently provided the lowest end-to-end delay among all protocols, making it a strong candidate for delay-sensitive applications, though its routing overhead can be higher.
- FSR's Niche: FSR showed better performance with higher traffic rates and in large-scale networks, but its throughput decreased with lower mobility, and it exhibited a higher routing load in certain scenarios.
- Mobility's Impact: Higher speeds and mobilities generally increase routing overhead for all protocols, as more route error messages are generated due to frequent link breakages.
Key Results
- Throughput Ranking (generally): DSR > AODV > DSDV > DYMO > OLSR > FSR.
- End-to-End Delay Ranking (generally): AODV > DSR > FSR > DSDV > DYMO > OLSR (OLSR had the lowest E2ED).
- Normalized Routing Load Ranking (generally): OLSR > DYMO > FSR > AODV > DSDV > DSR (DSR had the lowest NRL).
- Reactive protocols generally achieved higher throughput than proactive ones under high mobility.
- Proactive protocols generally had lower end-to-end delays than reactive ones in static or low-mobility scenarios.
Limitations and Future Work
Limitations: * The study relies solely on simulations using NS-2, and real-world performance might differ. * The authors acknowledge that the "route cache" mechanism in DSR and the "route table" in AODV are crucial for their performance, and the absence of a similar mechanism in DYMO contributes to its lower throughput. * FSR's performance is noted as "strange" in some scenarios, particularly its throughput decreasing with lower mobility, suggesting potential limitations in its route management for less dynamic environments.
Future Work: * The authors express interest in proposing and implementing a new ETX-based routing link metric with AODV and OLSR, recognizing the critical role of routing link metrics in protocol performance.
Summary updated