[Paper Review] A Utility Proportional Fairness Resource Allocation in Spectrally Radar-Coexistent Cellular Networks

Mo Ghorbanzadeh, Ahmed Abdelhadi+1 more|arXiv (Cornell University)|Jun 9, 2014

Wireless Communication Networks Research9 references4 citations

TL;DR

This paper proposes a two-stage convex optimization framework for utility proportional fairness resource allocation in spectrally coexistent cellular and radar networks. It first allocates radar spectrum to non-interfering sectors, then assigns remaining spectrum to all sectors, ensuring fairness and avoiding interference, with simulations showing balanced rate allocation even when radars are nearby.

ABSTRACT

Spectrum sharing is an elegant solution to addressing the scarcity of the bandwidth for wireless communications systems. This research studies the feasibility of sharing the spectrum between sectorized cellular systems and stationary radars interfering with certain sectors of the communications infrastructure. It also explores allocating optimal resources to mobile devices in order to provide with the quality of service for all running applications whilst growing the communications network spectrally coexistent with the radar systems. The rate allocation problem is formulated as two convex optimizations, where the radar-interfering sector assignments are extracted from the portion of the spectrum non-overlapping with the radar operating frequency. Such a double-stage resource allocation procedure inherits the fairness into the rate allocation scheme by first assigning the spectrally radar-overlapping resources.

Motivation & Objective

• Address the challenge of spectral coexistence between cellular networks and high-power radars to prevent mutual interference.
• Ensure quality of service (QoS) for diverse applications (real-time and delay-tolerant) in cellular networks sharing radar spectrum.
• Develop a resource allocation scheme that maintains fairness by compensating for initial resource exclusion of radar-interfering sectors.
• Formulate a double-stage optimization that assigns radar spectrum first to non-interfering sectors, then allocates remaining bandwidth to all sectors.

Proposed method

• Formulate the rate allocation problem as two convex optimization problems to ensure global optimality and fairness.
• Use sigmoidal and logarithmic utility functions to model delay-tolerant and real-time applications, respectively.
• Implement a two-stage allocation: Stage 1 assigns only radar-adjacent spectrum to non-interfering sectors (excluding interfering sectors), Stage 2 allocates non-overlapping communications spectrum to all sectors.
• Apply Lagrangian relaxation and dual decomposition techniques to solve the convex optimization problems efficiently.
• Incorporate radar interference constraints by excluding interfering sectors from radar-band allocation in the first stage.
• Ensure fairness by allocating more resources to interfering sectors in Stage 2 to compensate for their initial exclusion.

Experimental results

Research questions

• RQ1 How can spectrum be shared between cellular networks and radars without causing harmful interference?
• RQ2 How can resource allocation be optimized to maintain QoS for diverse application types in spectrally coexistent systems?
• RQ3 Can a two-stage allocation scheme ensure fairness while avoiding interference with radar operations?
• RQ4 What impact does radar proximity have on the achievable data rates of interfering cellular sectors?
• RQ5 How does the proposed method compare to non-coexistence allocation in terms of fairness and rate distribution?

Key findings

• The proposed two-stage scheme successfully avoids interference by allocating radar spectrum only to non-interfering sectors in the first stage.
• Interfering (red) sectors receive zero rate allocation in Stage 1, but experience a sharp rate increase in Stage 2 due to compensatory allocation.
• After Stage 2 begins, all sector rates grow closely together, indicating fair distribution despite initial exclusion.
• At a total rate of 250 units, the radar-interfering sector (R_comm^3) receives 50 bandwidth units from the communications spectrum, but only after a threshold of 200 units is reached.
• In contrast, without radar coexistence, the interfering sector receives bandwidth from the start, showing a fundamental difference in allocation behavior.
• The simulation results confirm that the scheme maintains fairness and QoS while ensuring spectral coexistence with radar systems.