Mobile jammers for secrecy rate maximization in cooperative networks

Dionysios S. Kalogerias, Nikolaos Chatzipanagiotis, Michael M. Zavlanos, Athina P. Petropulu

Abstract

This paper introduces a decentralized motion control scheme for mobile "helper" nodes that use cooperative jamming to protect wireless communications. By moving multi-antenna jammers to optimal locations and injecting "nulling noise" that affects an eavesdropper but not the intended receiver, the system can achieve high secrecy rates with significantly fewer nodes and less power than stationary setups. Numerical simulations validate that a single mobile jammer can match the performance of six stationary ones, demonstrating that mobility acts as a force multiplier for physical layer security.

Introduction

Wireless communication is inherently "leaky"—radio waves propagate in all directions, making it easy for an unauthorized eavesdropper (Eve) to intercept a message sent from a source (Alice) to a destination (Bob). While encryption handles this at the software level, Physical Layer Security (PLS) seeks to exploit the physics of the wireless channel itself to ensure Eve cannot decode the signal even with infinite computing power.

System model illustrating the interaction between Alice, Bob, Eve, and mobile helpers.

A common PLS technique is Cooperative Jamming (CJ), where "helper" nodes transmit artificial noise to drown out Alice’s signal for Eve. However, most existing CJ methods assume helpers are stationary or require perfect knowledge of Eve's location and channel. This paper addresses the challenge of coordinating mobile helpers to maximize the "secrecy rate"—the speed at which Alice can securely send data to Bob—without requiring a central controller or excessive power.

Methods

The proposed framework combines signal processing with robotic motion control through three main components:

  1. Nulling Noise (NN): Each helper node is equipped with multiple antennas. They use "Nulling Noise" to ensure that the jamming signal falls into the "null-space" of the channel to Bob. The noise is mathematically steered to cancel itself out at Bob’s location while remaining loud at Eve’s location.
  2. Decentralized Motion Control: The core innovation is a motion control law based on Artificial Potential Functions. Each helper calculates a "force" based on local channel conditions, pushing it toward positions that maximize the secrecy rate while avoiding collisions.
  3. Optimization: The problem is framed as jointly maximizing the secrecy rate by adjusting jamming weights and helper positions. Since nulling noise power is position-independent, the problem simplifies into a decentralized gradient-descent-like movement for each robot.

The algorithms were evaluated via numerical simulations in a 3x5 meter environment with a flat fading channel model (path-loss exponent of 3.5). Alice, Bob, and Eve were fixed, while 1 to 6 mobile helpers were initialized at various positions.

Key Results

The simulations revealed several critical insights regarding the efficacy of mobile jammers:

  • Mobility as a Force Multiplier: Moving a single jammer to an ideal spot is often more effective than adding more stationary jammers. As shown in Figure 2, a single mobile helper reached a secrecy rate nearly identical to the theoretical maximum ($R_{sup}$) after approximately 150 motion steps. Achieving a comparable rate with stationary helpers required at least six nodes.
  • Rapid Convergence: The secrecy rate typically starts at zero (if Eve initially has a better channel than Bob) and climbs rapidly as helpers move, reaching a steady state within a few dozen motion steps.
  • Power Efficiency: By optimizing their positions, helpers achieve the same security level using significantly less transmit power (lower Jamming Noise to Noise Ratio).
Secrecy rate as a function of motion steps for different numbers of mobile relays.

Discussion and Conclusion

The study demonstrates that decentralization is feasible; helpers do not need to coordinate with Alice or each other to design their noise, provided they can sense their own channel to Bob. While the motion control scheme guarantees local optimality, simulations show it consistently approaches the theoretical upper bound.

However, limitations remain. The model assumes helpers have a "map" of the channel to Eve, which may be impractical for passive eavesdroppers. Furthermore, the simple first-order motion model does not account for real-world robotic constraints like inertia. Future work should explore "pursuit-evasion" scenarios where helpers must track a moving Eve or maintain positions between a mobile Alice and Bob.

Summary updated

BibTeX
@inproceedings{kalogerias2013mobile,
  author = {Dionysios S. Kalogerias and Nikolaos Chatzipanagiotis and Michael M. Zavlanos and Athina P. Petropulu},
  title = {{Mobile jammers for secrecy rate maximization in cooperative networks}},
  booktitle = {IEEE International Conference on Acoustics, Speech and Signal Processing},
  year = {2013},
  url = {https://openreview.net/forum?id=jcRI3DZ1Qqj},
}
Figure