[Paper Review] Implementation of the SWIM Mobility Model in OMNeT++

Asanga Udugama, Behruz Khalilov+2 more|arXiv (Cornell University)|Sep 16, 2016

Opportunistic and Delay-Tolerant Networks3 references3 citations

TL;DR

This paper presents the implementation and validation of the Small Worlds in Motion (SWIM) mobility model in the OMNeT++ simulator, enabling realistic simulation of human mobility patterns in IoT networks. The model uses a probabilistic equation influenced by distance from home and location popularity to guide node movement, with verification showing that higher α values favor nearby locations while lower α values increase remote destination selection.

ABSTRACT

The Internet of Things (IoT) is expected to grow into billions of devices in the near future. Evaluating mechanisms such as networking architectures for communications in the IoT require the use of simulators due to the scale of the size of networks. Mobility is one of the key aspects that need to be evaluated when considering the different scenarios of the IoT. Small Worlds in Motion (SWIM) is a mobility model that mathematically characterises the movement patterns of humans. The work presented in this paper details the development and verification of the SWIM mobility model in the OMNeT++ simulator.

Motivation & Objective

• To address the gap in existing OMNeT++ mobility models that do not accurately reflect real human mobility patterns in IoT scenarios.
• To implement the SWIM mobility model—based on two core intuitions: proximity to home and location popularity—within the OMNeT++ simulation framework.
• To verify that the implemented model correctly reflects the theoretical behavior of SWIM, particularly the influence of the α parameter on destination selection.
• To provide a reusable, open-source implementation of SWIM in OMNeT++ for use in future IoT and mobile networking research.

Proposed method

• Extending the LineSegmentsMobilityBase class in OMNeT++ to implement the SWIM mobility logic, enabling segmented, linear movement with dynamic destination selection.
• Using a probabilistic destination selection mechanism based on the SWIM equation, which combines distance from home and location popularity to compute destination weights.
• Leveraging OMNeT++'s signal mechanism to notify nodes of arrivals at new locations, enabling dynamic updates to location popularity counts.
• Configuring model parameters such as α, waitTime, speed, and mobility area size to control movement behavior and simulate diverse mobility scenarios.
• Validating the model through simulation with varying α values (0.3 and 0.8) and analyzing resulting movement patterns in 2D space.
• Using a fixed set of 21 locations, with 16 classified as neighboring (within 300m of home) and 5 as remote, to assess selection bias under different α settings.

Experimental results

Research questions

• RQ1 How does the SWIM mobility model in OMNeT++ reproduce the theoretical behavior of human mobility patterns based on proximity and popularity?
• RQ2 To what extent does the α parameter in the SWIM model influence the selection of nearby versus remote destinations?
• RQ3 Can the SWIM model be effectively implemented and verified within the OMNeT++ simulation environment using standard mobility abstractions?
• RQ4 How do different α values affect the spatial distribution of node movements in a simulated IoT environment?

Key findings

• With α = 0.3, the model exhibits a higher tendency to select remote (visiting) locations, indicating that popularity plays a stronger role in destination choice.
• With α = 0.8, the model favors nearby (neighboring) locations, demonstrating that proximity to home dominates movement decisions.
• The simulation results confirm that the SWIM model correctly implements the theoretical trade-off between distance and popularity in destination selection.
• The implementation successfully integrates with OMNeT++'s mobility framework, using signals to propagate arrival events and update location popularity dynamically.
• The model’s behavior is consistent across multiple runs, showing stable and predictable movement patterns under defined parameter settings.
• The open-source implementation is available on GitHub under the GPL license, enabling reuse and extension by the research community.