Emergent Harmonic Field Routing for Dynamic Networks

Abstrakt

Routing in dynamic networks, such as MANETs,
remains a challenging problem due to the need for continuous
adaptation, robustness to topology changes, and limited coordination
among nodes. Traditional routing approaches, including
Distance Vector and Link-State protocols often underperform,
leading to instability or high overhead in dynamic environments.
In this paper, we propose Harmonic Field Routing (HFR), in
which forwarding decisions base on a self-organizing potential
field maintained through purely local interactions. Each node
iteratively updates a scalar state based on neighboring values,
leading to the implicit solution of a discrete Laplace equation
with Dirichlet boundary conditions. The resulting harmonic field
continuously adapts to network changes and induces a loop-free
routing structure, where packets follow the steepest descent of
the potential.
We analyze the theoretical foundations of the method by
relating it Jacobi iterative solver. Additionally, we propose two
asynchronous HFRs: Red-Black Gauss-Seidel HFR and Chaotic
HFR. Experimental evaluation shows that HFR methods provide
superior reliability in dynamic scenarios. Furthermore, due
to emergence of the early gradient, HFR can find the route
comparably fast to the Distance Vector.