This paper examines the impact of pilot contamination in multi-cell TDD systems, where uplink training using non-orthogonal sequences corrupts the channel estimates at base stations. The authors derive analytical expressions for achievable rates in a simplified setting and show that these rates can saturate with the number of base station antennas. They propose a new multi-cell MMSE-based precoding method to mitigate pilot contamination, which is linear and has a closed-form expression. Numerical results demonstrate significant performance gains over popular single-cell precoding methods. The paper emphasizes the importance of considering pilot contamination in multi-cell systems and suggests frequency/time reuse techniques to reduce its impact.This paper examines the impact of pilot contamination in multi-cell TDD systems, where uplink training using non-orthogonal sequences corrupts the channel estimates at base stations. The authors derive analytical expressions for achievable rates in a simplified setting and show that these rates can saturate with the number of base station antennas. They propose a new multi-cell MMSE-based precoding method to mitigate pilot contamination, which is linear and has a closed-form expression. Numerical results demonstrate significant performance gains over popular single-cell precoding methods. The paper emphasizes the importance of considering pilot contamination in multi-cell systems and suggests frequency/time reuse techniques to reduce its impact.