Description
Coarse Grain Scalability (CGS) is a layered video coding technique standardized within the 3GPP Multimedia Broadcast/Multicast Service (MBMS) framework, primarily specified in TS 26.904. It operates by structuring a video stream into multiple layers: a mandatory base layer and one or more enhancement layers. The base layer provides a basic quality video at a lower bitrate, while each successive enhancement layer incrementally improves the video quality, resolution, or frame rate when added. This layered approach is fundamental to scalable video coding (SVC), a profile of the H.264/AVC standard, where CGS represents one method of achieving spatial or quality scalability.
Architecturally, CGS is implemented within the video encoder and decoder. The encoder produces a bitstream where enhancement layers depend directly on the base layer or lower enhancement layers, but not on all preceding data, simplifying the decoding process. In the network, during transmission, layers can be selectively added or dropped based on real-time network conditions, device capabilities, or user subscription levels. For instance, in a congested cell, a network element like a broadcast multicast service center (BM-SC) might transmit only the base layer to ensure basic service continuity, whereas in a well-provisioned area, it could add enhancement layers to deliver high-definition content.
Key components involved include the SVC encoder, which generates the CGS layers; the transport mechanisms that packetize and deliver these layers, often using RTP/UDP/IP; and the client-side decoder that reconstructs the video from the received layers. The BM-SC plays a critical role in managing the distribution, potentially performing layer filtering or transcoding. CGS's role is to provide a robust, adaptable video delivery mechanism, particularly for broadcast and multicast services where a single stream must serve diverse receivers experiencing different channel conditions.
From a technical perspective, CGS works by applying quantization parameter (QP) offsets between layers. The base layer is encoded with a specific QP, and enhancement layers are encoded with finer quantization (smaller QP), adding detail. This differs from Medium Grain Scalability (MGS), which offers more granular quality adjustments using bit-plane coding. CGS layers are typically larger and provide coarser steps in quality improvement, making them efficient for scenarios where bandwidth variations are significant but not extremely fine-grained. Decoding requires that all lower layers be present; an enhancement layer cannot be decoded without its underlying base layer, ensuring dependency integrity.
In system operation, CGS enables dynamic adaptation without full re-encoding. Network nodes can truncate the bitstream by discarding enhancement layer packets, reducing bitrate instantly. This is vital for MBMS, such as evolved Multimedia Broadcast Multicast Service (eMBMS), where it helps manage radio resource allocation across multiple users. By supporting variable quality levels, CGS enhances user experience, network efficiency, and service flexibility, forming a foundational technology for adaptive streaming in mobile environments.
Purpose & Motivation
CGS was developed to address the challenges of delivering video content over mobile networks, which are characterized by heterogeneous devices and fluctuating bandwidth. Prior to scalable video coding, video streaming often used single-layer codecs like H.264/AVC, requiring separate encodings (transcoding) for different quality levels or network conditions. This approach was inefficient, increasing storage costs, computational overhead, and latency, especially for broadcast services where multiple bitrate variants had to be managed simultaneously. CGS, as part of SVC, introduced a more elegant solution by enabling a single encoded stream to adapt dynamically, reducing complexity and improving scalability.
The primary motivation for CGS stemmed from the growth of mobile video consumption and the need for efficient multicast/broadcast services in 3GPP networks. In Release 10, with enhancements to MBMS and the introduction of eMBMS in LTE, there was a demand for video coding that could seamlessly accommodate varying receiver capabilities—from smartphones with high-resolution displays to simpler devices with limited processing power—and diverse network conditions, such as cell-edge versus cell-center users. CGS provided a standardized method to achieve this, allowing operators to optimize bandwidth usage while maintaining service quality.
By solving these problems, CGS facilitated the deployment of services like mobile TV, live streaming, and software updates over multicast. It addressed limitations of previous non-scalable approaches by enabling graceful degradation in poor signal conditions and enhanced quality in good conditions, all from a single source stream. This not only improved user experience but also reduced operational costs for network operators, making video delivery more sustainable and flexible in the evolving 3GPP ecosystem.
Key Features
- Layered video encoding with base and enhancement layers
- Support for quality and spatial scalability adaptations
- Dynamic bitrate adjustment via layer addition/dropping
- Efficient single-stream encoding for multiple quality levels
- Compatibility with 3GPP MBMS and eMBMS frameworks
- Simplified decoding dependency structure compared to fine-grain scalability
Evolution Across Releases
Introduced CGS as part of scalable video coding enhancements for MBMS, based on H.264/SVC. It provided initial capabilities for coarse-grained quality layering to support adaptive video delivery in LTE broadcast services, enabling basic adaptation to network conditions and device diversity.
Defining Specifications
| Specification | Title |
|---|---|
| TS 26.904 | 3GPP TS 26.904 |