arXiv:2609.23657
QAOA-Based Pilot Assignment for Cell-Free Massive MIMO Systems
Xiaoyu Ma, Fang Fang, Xianbin Wang
Abstract
Pilot contamination is a major challenge in cell-free massive multiple-input multiple-output (MIMO) systems, where the limited number of orthogonal pilots makes pilot reuse unavoidable. Since the pilot assignment solution space grows exponentially with the number of users, finding the global optimum becomes computationally challenging on classical computers. Recent advances in quantum computing provide a promising approach for solving such large-scale combinatorial optimization problems. In this paper, we reformulate the assignment problem as a quantum-compatible optimization problem, enabling it to be directly solved by the Quantum Approximate Optimization Algorithm (QAOA). Specifically, the pilot contamination objective and assignment constraints are incorporated into the QAOA formulation. This allows the quantum search to focus on valid pilot assignments with low contamination cost. Simulation results show that the proposed method achieves performance close to exhaustive search, demonstrating the potential of quantum-assisted optimization for pilot assignment in future wireless systems.
AI Summary
The paper studies pilot assignment in cell-free massive MIMO, where users must reuse a limited set of orthogonal pilots. It minimizes a pairwise contamination surrogate, $C(\boldsymbol{\phi})=\sum_{j<k}\rho_{j,k}\mathbb{I}\{\phi_j=\phi_k\}$, with interference weights $\rho_{j,k}=\sum_m\beta_{m,j}\beta_{m,k}$ derived from large-scale fading. The assignment space has size $\tau_p^K$ for $K$ users and $\tau_p$ pilots. The proposed QAOA formulation uses one-hot encoding with $K\tau_p$ qubits. A diagonal cost Hamiltonian encodes the weighted pilot-sharing objective, and an all-pairs XY mixer within each user's register preserves the exactly-one-pilot constraint. Evolution starts from a uniform superposition of feasible assignments. Classical optimization minimizes the expected cost; the lowest-cost assignment among subsequent measurements is returned. Evaluation uses Qiskit's noiseless statevector simulator with 50 access points, three pilots, and $K=4,5,6,7$ users, averaged over 30 network realizations. Parameters are optimized using COBYLA with five random starts and at most 100 iterations, followed by 2,048 samples. At depth $p=3$, the reported sampled solutions approach exhaustive-search optima and modestly outperform greedy assignment. Expected-cost optimality gaps decrease substantially from $p=1$ to $p=2$, with smaller improvements thereafter and larger residual gaps at $K=7$; numerical gap values are not supplied in the text. The stated simulation-cost limitation restricts evaluation to seven users. Scalability improvements, noisy-device evaluation, and hardware validation are left for future work.
Central claim may not hold: The simulations cover only 81–2,187 feasible assignments under noiseless evolution and do not substantiate the motivating prospect of large-scale optimization. The near-optimality result selects the best of 2,048 QAOA samples without reporting a matched random-sampling baseline, although 2,048 uniform feasible samples would already find an optimum with probability at least 99.64% at the largest tested size. The supplied paper contains neither the referenced plots nor numerical performance results, preventing quantitative verification of its reported closeness to exhaustive search.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.