Channel Quality Indicator (CQI)

CQI is a value reported by the User Equipment (UE) to the base station (gNB/eNB), not measured directly by the network. The UE estimates downlink channel conditions from reference signals and maps them to a CQI index. This index corresponds to a recommended Modulation and Coding Scheme (MCS) that the UE believes it can decode with a Block Error Rate (BLER) not exceeding 10%.

• Reported on PUCCH (periodic) or PUSCH (aperiodic)
• Reporting periodicity configurable from 2ms to 160ms
• Wideband CQI reports a single value for the entire bandwidth
• Sub-band CQI provides per-sub-band granularity for frequency-selective scheduling

The 4-bit CQI value (0-15) maps directly to a specific modulation order and code rate. 5G NR defines three distinct CQI tables optimized for different spectral efficiency targets:

• Table 1 (64QAM max): Baseline table supporting up to 64QAM, suitable for standard mobile broadband
• Table 2 (256QAM max): Extended table for high-SINR scenarios, enabling 256QAM for peak throughput
• Table 3 (64QAM, low BLER): Ultra-reliable low-latency communication (URLLC) table targeting 10^-5 BLER

CQI 0 indicates out-of-range conditions where the UE cannot decode any transmission reliably.

The UE internally measures the Signal-to-Interference-plus-Noise Ratio (SINR) on Cell-Specific Reference Signals (CRS) or CSI-RS, then quantizes this continuous value into a discrete CQI index. This quantization process introduces a non-linear mapping that varies by UE implementation and chipset vendor.

• Typical SINR range for CQI 1-15: approximately -7 dB to +20 dB
• Each CQI step represents roughly 1-2 dB of SINR improvement
• UE receiver capabilities (e.g., MIMO detection algorithm) significantly influence the mapping
• Vendor-specific CQI compression algorithms may bias reports for proprietary scheduler optimization

CQI is a first-order feature in ML-based predictive load balancing and resource allocation models. Because CQI directly reflects the instantaneous channel quality, it serves as a leading indicator for future throughput. Predictive schedulers use historical CQI sequences to forecast:

• Future achievable spectral efficiency per UE
• Impending cell-edge conditions requiring proactive handover
• Multi-user MIMO pairing opportunities based on spatial channel correlation

A declining CQI trend often precedes a Radio Link Failure (RLF) by 100-500ms, providing a window for preemptive action.

Frequent, high-resolution CQI reporting improves scheduling accuracy but consumes precious uplink control resources. Network operators must balance:

• Periodicity: Shorter periods (e.g., 5ms) provide fresher channel state but increase PUCCH overhead
• Granularity: Sub-band reporting enables frequency-selective scheduling at the cost of larger payload sizes
• Wideband vs. Sub-band: Wideband CQI uses ~4 bits; sub-band CQI for 13 sub-bands requires ~52 bits per report

In massive MIMO systems with 64+ antennas, CSI-RS overhead for accurate CQI measurement becomes a significant design constraint.

CQI reporting is defined across multiple 3GPP releases with progressive enhancements:

• Release 8 (LTE): Introduced basic 4-bit CQI with 16 levels, wideband and UE-selected sub-band reporting
• Release 10 (LTE-Advanced): Added CSI-RS-based CQI for up to 8-layer transmission
• Release 15 (5G NR): Introduced three CQI tables, flexible CSI reporting settings, and beam-level CQI for mmWave
• Release 17: Enhanced CSI for high-velocity scenarios and AI/ML-based CSI compression study items

CQI is reported within the Channel State Information (CSI) framework alongside Rank Indicator (RI) and Precoding Matrix Indicator (PMI).

A comprehensive set of metrics reported by the UE that characterizes the downlink radio channel. While CQI is a specific index recommending a modulation and coding scheme, CSI is the broader framework that includes the Rank Indicator (RI) and Precoding Matrix Indicator (PMI). Together, these components enable the base station to perform adaptive modulation and coding (AMC) and MIMO beamforming. In 5G NR, CSI is measured using CSI-RS reference signals and reported periodically or aperiodically.

A table-driven index that directly determines the physical-layer data rate by specifying the modulation order (QPSK, 16QAM, 64QAM, 256QAM) and code rate. The gNB maps the reported CQI to an MCS using a vendor-specific link adaptation algorithm. A higher CQI enables a higher MCS, increasing throughput. Key considerations:

• Spectral efficiency: Higher MCS packs more bits per symbol
• BLER target: The MCS is chosen to maintain a target Block Error Rate, typically 10% for initial transmissions
• Outer loop link adaptation: Adjusts the CQI-to-MCS mapping based on HARQ feedback

The fundamental physical measurement from which CQI is derived. SINR represents the ratio of the desired signal power to the sum of interference and thermal noise. The UE measures SINR on CSI-RS or SSB reference signals and quantizes it into a CQI index. The relationship is non-linear and depends on UE receiver capability. Key points:

• Wideband SINR: Averaged across the entire carrier bandwidth
• Sub-band SINR: Reported per frequency sub-band for frequency-selective scheduling
• Interference sources: Neighboring cells, inter-cell interference, and thermal noise

The closed-loop process by which the base station dynamically adjusts transmission parameters based on UE channel reports. CQI is the primary input, but the algorithm also considers HARQ ACK/NACK statistics, buffer status, and QoS requirements. The goal is to maximize throughput while maintaining reliability. Two loops operate:

• Inner loop: Fast CQI-based MCS selection per TTI
• Outer loop: Slower adjustment of the SINR-to-CQI mapping offset based on BLER tracking This mechanism is critical for predictive load balancing as it directly influences per-user throughput.

The percentage of time-frequency resources allocated to user data transmission. CQI directly impacts PRB utilization efficiency: a low CQI requires more PRBs to transmit the same amount of data due to lower spectral efficiency. Predictive models use historical CQI and PRB utilization as multivariate time-series inputs to forecast future cell load. Key relationships:

• High CQI + low PRB usage: Efficient cell, capacity available
• Low CQI + high PRB usage: Congested cell, poor user experience
• CQI variance: High variance indicates mobility or interference instability

A 5G NR mechanism where the gNB triggers a one-shot CSI report using DCI format 0_1 rather than relying on periodic reporting. This enables on-demand CQI acquisition for specific scheduling decisions. Use cases:

• URLLC scheduling: Obtain fresh CQI before critical low-latency transmissions
• Beam management: Assess channel quality after a beam switch
• Predictive scheduling: Trigger CQI reports before predicted traffic bursts to optimize MCS selection Aperiodic reporting reduces uplink overhead while providing timely channel information for AI-driven resource allocation.