Optimal Filter Length and Zero Padding Length Design for Universal Filtered Multi-Carrier (UFMC) System

Lei Zhang, Ayesha Ijaz, Pei Xiao, Kezhi Wang, Deli Qiao, Muhammad Ali ImranIEEE Access 2019 · DOI

Introduction

Orthogonal Frequency Division Multiplexing (OFDM) has long been the standard for wireless communications, serving as the foundation for 4G LTE. However, as we move toward 5G and the Internet of Things (IoT), OFDM's high out-of-band (OoB) emissions have become a significant drawback, leading to spectral inefficiency. Universal Filtered Multi-Carrier (UFMC) has emerged as a promising alternative. By grouping subcarriers into subbands and applying filtering to each, UFMC significantly reduces spectral leakage.

Despite its advantages, UFMC introduces a complex design challenge: the "filtering headache." Selecting the appropriate filter and zero-padding lengths is critical. A filter that is too long introduces frequency selectivity that degrades the signal-to-noise ratio (SNR) and increases overhead. Conversely, a filter that is too short fails to protect against multipath fading and transceiver impairments like Carrier Frequency Offset (CFO) and Timing Offset (TO). This paper provides a rigorous mathematical framework to replace previous "rules of thumb" with optimal design criteria.

System Model and Methodology

The authors develop a comprehensive frequency-domain signal model for Multi-User (MU)-UFMC. This model explicitly accounts for hardware imperfections, including CFO, TO, and "Filter Tail Cutting"—a technique used to minimize system overhead.

Block diagram of the MU-UFMC system model including transmitter and receiver stages.

Central to their methodology is the derivation of Proposition 1, which defines the necessary and sufficient conditions for interference-free communication using one-tap equalization. The framework links FFT size and zero-padding directly to channel and filter lengths. Furthermore, the researchers introduce the Peak-to-Bottom-Power-Ratio (PBPR), a metric that quantifies the power "warping" across subcarriers caused by filtering, serving as a practical guide for filter selection.

The Filtering Trade-off: Shield vs. Penalty

A key insight of the paper is the dual nature of filtering in UFMC. In a perfectly synchronized system, UFMC can actually perform worse than OFDM. This is due to the "filtering penalty," where the filter response is uneven across subcarriers within a subband, as illustrated in the gain profiles for different filter lengths.

Filter gain across subcarriers for various filter lengths, illustrating the frequency selectivity introduced by UFMC.

However, the "filtering shield" becomes apparent when transceiver imperfections are introduced. In the presence of CFO and TO, UFMC’s subband filtering acts as a barrier, preventing interference from leaking across the entire bandwidth. This allows UFMC to maintain high Signal-to-Interference-plus-Noise Ratio (SINR) in scenarios where OFDM’s performance would collapse.

Optimization and Results

The paper formulates several optimization problems to maximize system capacity, ranging from fixed-overhead scenarios to "unequal" designs where each user receives a custom filter and padding length based on their specific channel conditions.

Key results include:

  • Robustness: In high-interference scenarios (e.g., significant CFO), the optimized UFMC system maintains significantly higher capacity than OFDM.
  • Equalization Gains: The proposed MMSE equalizers, which incorporate calculated interference power, outperform existing state-of-the-art receivers that assume perfect synchronization.
  • Design Guidelines: For channels with large delay spreads, such as Hilly Terrain (HT), the optimal zero-padding length is found to be much longer (~300 samples) than for simpler pedestrian channels.

Conclusion and Future Directions

This research provides the analytical tools needed to transition UFMC from a theoretical candidate to a practical 5G/IoT waveform. While the current work focuses on the downlink and Chebyshev filters, future research could explore the higher complexity of multi-user interference in the uplink and lower-complexity heuristic algorithms for real-time optimization in systems with many subbands.

Summary updated

BibTeX
@article{zhang2019optimal,
  author = {Lei Zhang and Ayesha Ijaz and Pei Xiao and Kezhi Wang and Deli Qiao and Muhammad Ali Imran},
  title = {{Optimal Filter Length and Zero Padding Length Design for Universal Filtered Multi-Carrier (UFMC) System}},
  journal = {IEEE Access},
  year = {2019},
  doi = {10.1109/access.2019.2898322},
}
Figure