Connectivity vs. control: Using directional and positional cues to stabilize routing in robot networks

Karthik Dantu, Gaurav S. Sukhatme

Abstract

In multi-robot systems, coordination depends on a stable communication graph. However, standard mobile ad-hoc network (MANET) protocols often prioritize the "shortest path," which can be highly fragile in mobile environments. This paper proposes augmenting the Optimized Link State Routing (OLSR) protocol with directional and positional cues to predict link longevity. By prioritizing topological persistence over hop-count optimality, the proposed method reduces disruptive route switches by up to 20% in simulated environments.

Introduction

In applications such as drone swarms or search-and-rescue teams, robots must maintain constant communication to coordinate effectively. Most MANET protocols are "mobility-unaware," optimizing for the minimum number of hops between nodes. This often leads to "route flapping," where the protocol frequently switches paths because the shortest link—often between robots moving in opposite directions—breaks quickly. For control algorithms, this instability introduces significant noise. This research argues that a slightly longer, more stable path is superior to a perfect path that breaks in seconds.

Methodology

The researchers modified the Optimized Link State Routing (OLSR) protocol, a proactive protocol that periodically maps network topology, by introducing two types of mobility-aware cues.

Directional Cues

Robots share their current velocity vectors with one-hop neighbors. Neighbors calculate the angular offset between their own motion and the sender's. This offset is mapped to a stability metric ranging from 1 to 8.

Mapping angular offset to a stability metric (1–8) based on relative motion.

As shown in the figure above, links between robots moving in the same direction are assigned higher stability values, while those moving apart are deemed unstable.

Positional Cues and Protocol Augmentation

By sharing $(x, y)$ coordinates and velocity, robots can estimate the "Link Duration"—the time remaining before they exit each other's communication range, based on a simple disc radio model. To implement this, the standard OLSR "Hello" messages were augmented with positional and directional tuples.

Network model illustrating communication range and the augmented OLSR message formats for Direction and Location cues.

The protocol logic was modified to select relays with the highest stability metric when choosing paths to 2-hop neighbors. Additionally, a hysteresis mechanism was introduced to prevent unnecessary jumping; the protocol only switches an existing route if its stability falls below a specific threshold.

Results and Discussion

The proposed modifications were evaluated using ns-2 simulations in a 500m x 500m area with varying node densities and speeds.

  • Reduced Route Flapping: Directional cues alone reduced route switches by 2–8%. When combined with positional data, the system achieved up to a 20% reduction in switches compared to baseline OLSR.
  • Density and Speed: Benefits were more pronounced in denser networks due to the availability of more redundant paths. However, as robot speed increased, the predictive benefit diminished because links broke too frequently for any cueing to maintain stability.
  • Error Sensitivity: The system proved robust to positional errors (e.g., GPS inaccuracies) but was highly sensitive to directional errors. Incorrect heading reports caused stability predictions to fail rapidly.

Conclusion

By treating mobility as a predictable feature rather than a random failure, robot networks can achieve significantly higher stability. Future work will focus on moving beyond the simplified disc radio model to account for multi-path fading and obstacles, as well as testing these algorithms on physical robot platforms.

Summary updated

BibTeX
@inproceedings{dantu2009connectivity,
  author = {Karthik Dantu and Gaurav S. Sukhatme},
  title = {{Connectivity vs. control: Using directional and positional cues to stabilize routing in robot networks}},
  booktitle = {ROBOCOMM},
  year = {2009},
  doi = {10.4108/icst.robocomm2009.5846},
}
Figure