Quality of Service (QoS)

The actual rate at which data is successfully transmitted over a communication channel in a given period, typically measured in bits per second (bps). Unlike raw bandwidth, throughput accounts for real-world protocol overhead, retransmissions, and network congestion.

• Peak Throughput: The maximum achievable rate under ideal conditions, as defined in 3GPP specifications for enhanced mobile broadband (eMBB).
• Cell-Edge Throughput: The data rate experienced by users at the boundary of a cell, often the 5th percentile user, critical for fairness guarantees.
• Application-Layer Throughput: The effective rate perceived by the end-user application, excluding TCP/IP and radio protocol overhead.

The time delay between a data packet's transmission and its successful reception, measured in milliseconds (ms). It is a critical constraint for real-time applications and is decomposed into multiple components within a network.

• Radio Access Network (RAN) Latency: The one-way transit time over the air interface, targeted at <1 ms for Ultra-Reliable Low-Latency Communication (URLLC) in 5G.
• Round-Trip Time (RTT): The total time for a signal to travel to a destination and back, including processing delays.
• End-to-End (E2E) Latency: The cumulative delay from the user equipment through the core network to the application server.

The statistical variance in the arrival time of successive data packets, measured in milliseconds. High jitter causes buffer underruns and overruns, severely degrading the quality of real-time voice and video communication.

• Packet Delay Variation (PDV): The formal metric defined in IETF RFC 3393 for quantifying the difference in one-way delay between selected packets.
• De-Jitter Buffering: A receiver-side technique that intentionally adds a small delay to smooth out arrival time variations, trading increased latency for reduced jitter.
• Jitter Tolerance: The maximum PDV an application can accept before experiencing artifacts; for conversational voice, this is typically < 30 ms.

The fraction of transmitted data packets that fail to reach their intended destination, expressed as a percentage. Loss occurs due to channel errors, congestion-induced buffer overflows, or handover failures.

• Block Error Rate (BLER): The radio-layer metric for transport block decoding failures, typically maintained at 10% for initial transmissions via link adaptation.
• Packet Error Loss Rate (PELR): The upper-layer QoS parameter defined by 3GPP, with URLLC services requiring a PELR of 10^-5 or better.
• Retransmission Overhead: The capacity cost incurred by Automatic Repeat Request (ARQ) and Hybrid ARQ (HARQ) mechanisms used to recover lost packets.

A subjective quality metric, expressed on a scale from 1 (bad) to 5 (excellent), that quantifies the human user's perceived quality of a voice or video stream. While subjective by origin, it is often estimated algorithmically from network QoS metrics.

• PESQ/POLQA: Perceptual Evaluation of Speech/Video Quality algorithms that map objective signal measurements to predicted MOS values.
• E-Model: An ITU-T G.107 computational model that estimates conversational quality from transmission parameters including delay, echo, and codec distortion.
• QoE vs. QoS: Quality of Experience (QoE) is the user-centric superset of QoS; a network can have excellent QoS metrics but poor QoE due to application-layer stalling.