Opportunistic Spectrum Allocation for Max-Min Rate in NC-OFDMA

TL;DR

This paper proposes a spectrum allocation method for NC-OFDMA cognitive radio links that maximizes the minimum rate while limiting spectral span, balancing performance and power consumption, validated through optimization and experimental testing.

Contribution

It introduces a max-min rate spectrum allocation framework for NC-OFDMA links that considers spectral span constraints, providing an efficient solution and demonstrating practical effectiveness.

Findings

The proposed allocation achieves near tradeoff boundary performance.

Limiting spectral span incurs minimal rate penalty.

The method is validated through USRP testbed experiments.

Abstract

We envision a scenario of opportunistic spectrum access among multiple links when the available spectrum is not contiguous due to the presence of external interference sources. Non-contiguous Orthogonal Frequency Division Multiplexing (NC-OFDM) is a promising technique to utilize such disjoint frequency bands in an efficient manner. In this paper we study the problem of fair spectrum allocation across multiple NCOFDM-enabled point-to-point cognitive radio links under certain practical considerations that arise from such non-contiguous access. When using NC-OFDMA, the channels allocated to a cognitive link are spread across several disjoint frequency bands leading to a large spectral span for that link. Increased spectral span requires higher sampling rates, leading to increased power consumption in the ADC/DAC of the transmit/receive nodes. In this context, this paper proposes a spectrum allocation that maximizes the minimum rate achieved by the cognitive radio links, under a constraint on the maximum permissible spectral span. Under constant transmit powers and orthogonal spectrum allocation, such an optimization is a mixed-integer linear program and can be solved efficiently. There exists a clear trade-off between the max-min rate achieved and the maximum permissible spectral span. The spectral allocation obtained from the proposed optimization framework is shown to be close to the tradeoff boundary, thus showing the effectiveness of the proposed technique. We find that it is possible to limit the spectrum span without incurring a significant penalty on the max-min rate under different interference environments. We also discuss an experimental evaluation of the techniques developed here using the Universal Software Radio Peripheral (USRP) enabled ORBIT radio network testbed.