Sounding Reference Signal

Description

The Sounding Reference Signal (SRS) is a known sequence transmitted by the UE on the uplink physical layer. Unlike data or control signals, its primary purpose is not to carry information but to 'sound' the channel. The gNB (in NR) or eNB (in LTE) uses the received SRS to estimate the uplink channel state information (CSI), including characteristics like frequency response and spatial properties. This estimation is performed over the bandwidth configured for the SRS transmission, which can be wideband (covering a large portion of the carrier) or narrowband (a specific subband).

The transmission of SRS is configurable by the network via RRC signaling and/or MAC CE and DCI. Key configurable parameters include periodicity, time/frequency resource allocation (comb structure and bandwidth), sequence cyclic shift, and transmission comb. A crucial feature introduced in LTE-Advanced and enhanced in NR is antenna switching SRS, where a UE with multiple transmit antennas can sound using different antennas across different SRS symbols or subframes, allowing the base station to learn the uplink channel for each antenna port. This supports uplink transmit diversity and MIMO. In NR, SRS functionality has been significantly expanded. It supports flexible positioning across the OFDM symbol (front-loaded or rear-loaded), wider bandwidths, and multi-symbol transmission within a slot. NR also introduced the concept of SRS resource sets, allowing for grouped configurations for different purposes like codebook-based transmission, non-codebook-based transmission, antenna switching, and beam management.

How it works: The network configures the UE with SRS resources. At the configured time, the UE generates the SRS sequence based on its parameters and transmits it. The gNB receives this known sequence and, by comparing it with the locally generated version, can estimate how the uplink channel has altered the signal. This channel estimate is then used for several critical functions: 1) Uplink channel-aware scheduling, allocating resources to UEs with favorable channel conditions; 2) Uplink link adaptation, selecting the appropriate MCS; 3) For frequency-selective scheduling, determining which subbands are best for a particular UE; and 4) For uplink multi-user MIMO (UL-MU-MIMO), determining which UEs can be spatially multiplexed on the same resources without excessive interference. In NR, SRS also plays a vital role in beam management for uplink beamforming, especially in FR2 (mmWave) bands.

Purpose & Motivation

The SRS was created to address a fundamental asymmetry in early LTE design: while the downlink had dedicated reference signals (CRS) for channel estimation, the uplink lacked a dedicated, UE-specific channel sounding mechanism for wideband or spatial channel estimation. Initial LTE (Rel-8) relied on demodulation reference signals (DM-RS) associated with PUSCH for data decoding, but these only provided channel estimates for the scheduled bandwidth, limiting the network's knowledge for future scheduling decisions and advanced antenna techniques.

The introduction of SRS in LTE Rel-9 solved this problem by providing a configurable, network-controlled mechanism for the base station to proactively learn the uplink channel state. This enabled uplink frequency-selective scheduling, improving spectral efficiency. The primary motivation evolved with LTE-Advanced and NR: to support uplink MIMO and multi-antenna techniques. Without SRS, the base station cannot accurately estimate the spatial characteristics of the uplink channel from multiple UE antennas, which is essential for precoding, spatial multiplexing, and beamforming. SRS thus became the enabler for advanced features like uplink multi-user MIMO (MU-MIMO), where multiple UEs share the same time-frequency resources, and for efficient operation in higher frequency bands where beam-based operation is critical. It transformed the uplink from a largely blind scheduling process to a channel-aware, optimized one.