Near-optimal pilot assignment in cell-free massive MIMO

Raphael M. Guedes,José F. de Rezende,Valmir C. Barbosa

2024-10-01

Abstract:Cell-free massive MIMO systems are currently being considered as potential enablers of future (6G) technologies for wireless communications. By combining distributed processing and massive MIMO, they are expected to deliver improved user coverage and efficiency. A possible source of performance degradation in such systems is pilot contamination, which contributes to causing interference during uplink training and affects channel estimation negatively. Contamination occurs when the same pilot sequence is assigned to more than one user. This is in general inevitable, as the number of mutually orthogonal pilot sequences corresponds to only a fraction of the coherence interval. We introduce a new algorithm for pilot assignment and analyze its performance both from a theoretical perspective and in computational experiments. We show that it has an approximation ratio close to 1 for a plausibly large number of orthogonal pilot sequences, as well as low computational complexity under massive parallelism. We also show that, on average, it outperforms other methods in terms of per-user SINR and throughput on the uplink.

Information Theory,Networking and Internet Architecture

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to effectively allocate pilot sequences in cell - free massive multiple - input multiple - output (massive MIMO) systems to reduce the performance degradation caused by pilot contamination. Specifically, when multiple users use the same pilot sequence, it will lead to an increase in interference during uplink training, thus affecting the accuracy of channel estimation. This contamination is particularly serious when the number of pilot sequences is less than the number of users, because at this time multiple users must share the same pilot sequence.

To meet this challenge, the paper proposes a new algorithm - the Greedy Edge Contraction (GEC) algorithm for optimizing pilot allocation. This algorithm is achieved by solving the MAX P - CUT problem, where P represents the number of available orthogonal pilot sequences. The goal of the GEC algorithm is to minimize the sum of the interference variances caused by pilot contamination, thereby increasing the signal - to - interference - plus - noise ratio (SINR) and uplink throughput for each user.

The paper also analyzes the theoretical performance of the GEC algorithm and verifies its effectiveness and superiority in practical applications through computational experiments. The results show that the GEC algorithm is superior to other existing pilot allocation methods in most cases, especially when the number of pilot sequences is large, and its performance is close to the optimal solution.

### Formula Summary
- **Channel Model**:
\[
g_{mk}=\beta_{mk}^{1 / 2}h_{mk}
\]
where $\beta_{mk}$ represents large - scale fading and $h_{mk}$ represents small - scale fading.

- **Interference Variance**:
\[
v_{k}=\sum_{k^{\prime} \in U_{k} \setminus \{k\}} \sum_{m = 1}^{M}\beta_{mk^{\prime}}
\]

- **Objective Function**:
\[
\text{Minimize}\quad\sum_{S \in P}\sum_{k \in S}v_{k}
\]
where $P = \{S_{1},\ldots,S_{P}\}$ is a partition of the user set, and users in each subset $S$ are allocated the same pilot.

- **Edge Weights in the GEC Algorithm**:
\[
w_{ij}=\beta_{k}+\beta_{k^{\prime}}
\]
where $\beta_{k}=\sum_{m = 1}^{M}\beta_{mk}$.

- **Time Complexity of the GEC Algorithm**:
- Time complexity on sequential devices: $O(K^{3})$
- Time complexity on massively parallel devices: $O(K\log K)$

Through these formulas and algorithm designs, the paper provides an effective solution to reduce the impact of pilot contamination on the performance of cell - free massive MIMO systems.