Optimal Frequency Band Allocation for Multi-Cell IoT Private Networks over Unlicensed Spectrum

The optimization of frequency reuse of multi-cell Internet of Things (IoT) private networks over an unlicensed spectrum is crucial for effectively supporting massive IoT devices but has yet to be studied. This optimization is more challenging than the counterpart for multi-cell cellular networks over licensed spectrum since the additional influences of varying interference over unlicensed spectrum, activities of IoT devices, and ALOHA random access should be carefully considered. This paper investigates optimal frequency band allocation under frequency reuse for a multi-cell IoT private network operating on an unlicensed wide band and adopting ALOHA random access. The frequency band allocation is quasistatic, adapts only to the network statistics, and considers the frequency band adjustment costs, making it easy to implement in practice. The optimization of frequency band allocation is formulated as a non-differentiable integer program (IP), converted to an equivalent linear program (LP) under certain conditions, and solved efficiently using the partial dual decomposition method. Analytical and numerical results offer design insight and validate the significant value of the proposed solution.