[Paper Review] Performance Comparison of the RPL and LOADng Routing Protocols in a Home Automation Scenario

Mališa Vučinić, Bernard Tourancheau+1 more|arXiv (Cornell University)|Jan 6, 2014

Energy Efficient Wireless Sensor Networks10 references4 citations

TL;DR

This paper evaluates RPL and LOADng in a realistic home automation scenario using Contiki OS and Cooja simulation, finding that RPL outperforms LOADng in delay, control overhead, and memory usage despite its higher complexity. LOADng's performance is highly sensitive to Route Hold Time, and optimization of flooding and route storage could improve its scalability.

ABSTRACT

RPL, the routing protocol proposed by IETF for IPv6/6LoWPAN Low Power and Lossy Networks has significant complexity. Another protocol called LOADng, a lightweight variant of AODV, emerges as an alternative solution. In this paper, we compare the performance of the two protocols in a Home Automation scenario with heterogenous traffic patterns including a mix of multipoint-to-point and point-to-multipoint routes in realistic dense non-uniform network topologies. We use Contiki OS and Cooja simulator to evaluate the behavior of the ContikiRPL implementation and a basic non-optimized implementation of LOADng. Unlike previous studies, our results show that RPL provides shorter delays, less control overhead, and requires less memory than LOADng. Nevertheless, enhancing LOADng with more efficient flooding and a better route storage algorithm may improve its performance.

Motivation & Objective

• To evaluate and compare the performance of RPL and LOADng in a realistic home automation scenario with mixed multipoint-to-point and point-to-multipoint traffic.
• To analyze the impact of network density and traffic patterns on routing protocol efficiency in constrained LLNs.
• To assess the trade-offs between proactive (RPL) and reactive (LOADng) routing approaches in terms of delay, overhead, and memory usage.
• To identify implementation-level bottlenecks in LOADng, particularly related to flooding and route storage, and suggest optimization directions.
• To provide empirical insights into protocol suitability for real-world home automation deployments with latency and energy constraints.

Proposed method

• Simulation of a realistic home automation network topology using Contiki OS and the Cooja simulator/emulator.
• Implementation and evaluation of ContikiRPL (RPL) and a basic, non-optimized version of LOADng.
• Use of a Unit Disk Graph radio propagation model to simulate realistic wireless conditions.
• Simulation of heterogeneous traffic patterns including multipoint-to-point and point-to-multipoint flows.
• Evaluation of key performance metrics: end-to-end delay, control plane overhead, number of hops, and memory usage.
• Parameter tuning of Route Hold Time (RHT) in LOADng to study its impact on flooding frequency and overhead.

Experimental results

Research questions

• RQ1 How do RPL and LOADng compare in terms of end-to-end delay under mixed traffic patterns in a dense home automation network?
• RQ2 What is the control plane overhead of RPL and LOADng, and how does it scale with increasing network density and RHT in LOADng?
• RQ3 How do memory requirements and route storage mechanisms differ between RPL and LOADng in a constrained device environment?
• RQ4 To what extent does the choice of Route Hold Time in LOADng affect its performance and scalability in dense deployments?
• RQ5 Can optimizing flooding and route storage in LOADng close the performance gap with RPL in real-world home automation scenarios?

Key findings

• RPL achieves significantly lower end-to-end delays than LOADng, particularly in dense network topologies with mixed traffic.
• RPL exhibits lower control plane overhead compared to LOADng, especially when the Route Hold Time in LOADng is short.
• RPL requires less memory than LOADng due to its more efficient route storage and proactive route maintenance.
• LOADng’s control overhead increases sharply with network density and decreases only with longer Route Hold Times, indicating poor scalability in dense smart home environments.
• The performance of RPL is highly dependent on implementation details and parameter tuning, particularly for DAO and DIO message intervals.
• Optimizing flooding and route storage in LOADng could reduce its overhead and improve its viability for dense, latency-sensitive applications.