[Paper Review] Optimization of an Adaptive Frequency-Hopping Network

Salvatore Talarico, Matthew C. Valenti+1 more|arXiv (Cornell University)|Aug 24, 2015

Wireless Communication Networks Research10 references3 citations

TL;DR

This paper proposes an optimization framework for adaptive frequency-hopping networks using continuous-phase frequency-shift keying (CPFSK) and rate-adaptive coding to maximize area spectral efficiency under co-channel and adjacent-channel interference. By jointly optimizing the modulation index, number of frequency channels, and fractional in-band power while adapting code rate per network realization, the method improves normalized modulation-constrained area spectral efficiency (MASE) by over 200% compared to arbitrary parameter selection.

ABSTRACT

This paper proposes a methodology for optimizing a frequency-hopping network that uses continuous-phase frequency-shift keying and adaptive capacity-approaching channel coding. The optimization takes into account the spatial distribution of the interfering mobiles, Nakagami fading, and lognormal shadowing. It includes the effects of both co-channel interference and adjacent-channel interference, which arises due to spectral-splatter effects. The average network performance depends on the choice of the modulation index, the number of frequency-hopping channels, and the fractional in-band power, which are assumed to be fixed network parameters. The performance of a given transmission depends on the code rate, which is adapted in response to the interference to meet a constraint on outage probability. The optimization proceeds by choosing a set of fixed network parameters, drawing the interferers from the spatial distribution, and determining the maximum rate that satisfies the outage constraint. The process is repeated for a large number of network realizations, and the fixed network parameters that maximize the area spectral efficiency are identified.

Motivation & Objective

• Address the performance degradation in ad hoc frequency-hopping networks due to co-channel and adjacent-channel interference.
• Overcome the limitations of fixed network parameters (modulation index, channel count, in-band power) that lead to variable outage probability under dynamic interference.
• Develop a systematic optimization method to maximize area spectral efficiency (MASE) under realistic propagation conditions including Nakagami fading and lognormal shadowing.
• Enable dynamic rate adaptation per network realization to maintain a target outage probability while improving fairness and throughput.
• Quantify the tradeoffs between spectral efficiency, interference, and system parameters to identify globally optimal configurations.

Proposed method

• Model the spatial distribution of interferers using a Poisson point process to represent mobile density in ad hoc networks.
• Use a physical interference model that accounts for both co-channel interference (CCI) and adjacent-channel interference (ACI) due to spectral splatter.
• Derive a closed-form expression for the conditional outage probability based on network geometry and channel conditions (Rayleigh, Nakagami, or mixed fading).
• Adapt the code rate for each network realization to satisfy a target outage probability constraint, ensuring reliable communication under varying interference levels.
• Compute the average area spectral efficiency (MASE) by averaging over spatial and fading distributions, with rate adaptation performed per network instance.
• Perform Monte Carlo simulations over numerous network realizations to identify the optimal fixed parameters: modulation index $h$, number of channels $L$, and fractional in-band power $\psi$ that maximize normalized MASE.

Experimental results

Research questions

• RQ1 How does adjacent-channel interference (ACI) impact the performance of frequency-hopping networks when spectral splatter is present?
• RQ2 What is the optimal tradeoff between bandwidth efficiency and error performance when adjusting the modulation index $h$ in CPFSK-based FH systems?
• RQ3 How does reducing the fractional in-band power (e.g., from 99% to 95%) affect system throughput despite increased ACI?
• RQ4 What is the impact of network density and fading environment (Rayleigh, Nakagami, mixed, shadowed) on the optimal configuration of $L$, $h$, and $\psi$?
• RQ5 To what extent can rate adaptation improve fairness and spectral efficiency compared to fixed-rate systems under dynamic interference?

Key findings

• Optimizing the modulation index $h$, number of frequency channels $L$, and fractional in-band power $\psi$ can improve normalized MASE by more than 200% compared to a standard configuration $(L=200, h=0.5, \psi=0.99)$.
• For a dense network with 50 interferers, the optimal configuration achieves a normalized MASE of 5.00 bps/kHz·m² under mixed fading (LOS for direct link, Rayleigh for interferers), compared to 3.56 for the standard choice.
• Reducing fractional in-band power to 95% improves spectral efficiency by enabling more frequency channels or lower-rate, more robust codes, even with increased ACI.
• The optimal parameters are robust across different fading models: $h \approx 0.80$ and $\psi = 0.96$ are consistently optimal, while $L$ increases with interferer density.
• Shadowing degrades performance significantly, requiring higher $L$ and $\psi$ to maintain MASE, and reduces optimal MASE by up to 40% compared to unshadowed scenarios.
• The CDF of the adapted code rate is steeper in dense networks, indicating improved fairness in rate allocation due to more predictable interference levels.