A multi-stage approach for Root Sequence Index allocation
Abstract
The Root Sequence Index (RSI) is a parameter used in mobile wireless networks to allocate uplink channels between user equipment and base stations. Inadequate RSI assignment to neighbor radios may lead to failure in service establishment and performance degradation. Wireless networks are also dynamic, with uncertain modifications in time due to, for instance, seasonal foliage. In this paper, we model RSI allocation with seasonal changes in the network as a multistage robust problem, being the first proactive, look-ahead method to consider uncertainty in RSI assignments. We develop methods to solve this stochastic problem, aiming to minimize the possible interference and network changes in time. A mixed-integer programming model, a classic Biased Random-Key Genetic Algorithm (BRKGA), and a novel BRKGA hybridized with Dijkstra’s algorithm are explored and compared. We also introduce a Monte Carlo-based simulation methodology to obtain scenarios. The hybrid BRKGA-based approach is shown to obtain more robust solutions in shorter computational times.