Discrete Fourier Transform

Description

The Discrete Fourier Transform (DFT) is a cornerstone signal processing technique within the 3GPP physical layer, specifically for generating and demodulating OFDM and SC-FDMA waveforms. In the transmitter, the DFT transforms a block of complex modulation symbols from the time domain into the frequency domain. For uplink SC-FDMA in LTE, this process is implemented via a DFT-precoding stage before mapping to subcarriers, which reduces the Peak-to-Average Power Ratio (PAPR) and improves power amplifier efficiency in User Equipment (UE). In the downlink OFDM for both LTE and 5G NR, an Inverse DFT (IDFT) is used at the transmitter to convert frequency-domain mapped symbols into a time-domain OFDM symbol for transmission over the air interface.

Architecturally, the DFT operation is integrated into the baseband processing chain of both the gNB/NodeB (base station) and the UE. Its implementation is often optimized as a Fast Fourier Transform (FFT), a computationally efficient algorithm for calculating the DFT. The size of the DFT, often referred to as the FFT size, is a key parameter that determines the number of subcarriers and thus the system bandwidth. For instance, in LTE, standard FFT sizes like 128, 256, 512, 1024, and 2048 correspond to different channel bandwidths from 1.4 MHz to 20 MHz. In 5G NR, the concept is extended with more flexible numerology, supporting a wider range of subcarrier spacings and corresponding DFT sizes to cater to diverse use cases from enhanced mobile broadband to ultra-reliable low-latency communications.

The DFT's role is pivotal for enabling orthogonal multiplexing of multiple users and data streams. By converting signals to the frequency domain, the system can precisely allocate specific subcarriers to different users (as in OFDMA), ensuring orthogonality and minimizing inter-carrier interference. At the receiver, the DFT is applied again to convert the received time-domain signal back to the frequency domain, allowing for channel estimation, equalization, and demodulation of the transmitted symbols. This transformation is essential for dealing with frequency-selective fading in wireless channels, as it allows for per-subcarrier equalization, significantly simplifying the receiver design compared to time-domain equalizers for wideband signals.

Furthermore, the DFT underpins critical physical layer procedures. It is used in the generation of reference signals (like DM-RS and SRS) for channel estimation and in the implementation of advanced multi-antenna techniques such as precoding for MIMO. The efficiency and accuracy of the DFT/FFT implementation directly impact system performance metrics like throughput, latency, and power consumption. Its computational requirements also influence UE and base station hardware design, making it a key consideration for silicon implementation.

Purpose & Motivation

The DFT was incorporated into 3GPP standards starting with LTE (Rel-8) to address the limitations of the Wideband Code Division Multiple Access (WCDMA) used in 3G UMTS. WCDMA, a direct-sequence spread spectrum technology, faced challenges with high Peak-to-Average Power Ratio (PAPR) and complexity in equalizing wideband signals over frequency-selective channels. These factors limited spectral efficiency and increased power amplifier cost and drain in mobile devices. The adoption of OFDM-based access schemes (OFDMA downlink, SC-FDMA uplink) necessitated the DFT as the core mathematical engine to enable efficient multi-carrier transmission.

The primary problem the DFT solves is the efficient and orthogonal separation of signals in the frequency domain. By using the DFT, 3GPP systems can divide the available spectrum into numerous narrowband, orthogonal subcarriers. This orthogonality prevents inter-symbol interference within a symbol period and allows for flexible spectrum allocation. For the uplink, the specific use of DFT-precoding in SC-FDMA was motivated by the need for a low-PAPR waveform to extend UE battery life and reduce the cost and linearity requirements of the UE power amplifier, a critical advantage over pure OFDMA.

Historically, the motivation stemmed from the industry's drive towards higher data rates and improved spectral efficiency for 4G and beyond. The DFT, through its FFT implementation, provided a computationally feasible method to handle the large bandwidths required for high-speed data. Its integration allowed LTE and subsequently 5G NR to support scalable bandwidths, advanced antenna technologies, and robust performance in challenging multipath environments, forming the mathematical foundation for the physical layer that defines modern cellular broadband.