Don’t Lag Behind: Understand AVoIP Latency
Kevin Wang
VP, MXnet Systems Group
Latency is one of the most critical factors when designing and implementing AV over IP systems, and understanding it is crucial for various stakeholders. This article will explore the concept of latency in AV over IP applications, define key terms, examine different types of latency and their relevance across vertical industries, and explain how latency can be tested in a real-world setup.
What is Latency?
In the context of AV over IP solutions, latency refers to the delay between the time a signal leaves the source (e.g., a video camera or a media player) and the time it appears on the display device after traveling through the encoder, network, decoder, and eventually the display. In simple terms, latency measures how long it takes for the video signal to make the journey from source to screen. It is often expressed in milliseconds (ms) or microseconds (μs).
Common latency descriptors in AV over IP solutions include:
- Low Latency: Typically, under 50 ms, approximately equal to the duration of one video frame.
- Ultra-Low Latency: Generally, refers to sub-frame latency, meaning a delay of just a few milliseconds (ms) — lower than one frame.
- Zero Latency: Less than 100 microseconds (μs) (0.1 ms).
These descriptors are not arbitrary but have a direct relationship with the type of video codec used and the level of compression applied to the video signal.
Compression Algorithms and Their Effect on Latency
Compression is a key factor affecting latency. There is a delicate balance between compression ratio, image quality, and the latency introduced during the encoding and decoding process:
- High Compression Ratio Codecs: These codecs, such as H.264/265, are often used when Low Latency is acceptable, and efficient bandwidth use is a priority. High compression leads to reduced data size, but the complex encoding process may introduce more latency.
- Moderate Compression Ratio Codecs: Codecs like JPEG2000 and MJPEG offer a middle ground and are commonly used for Ultra-Low Latency solutions. They offer less aggressive compression compared to H.264/265, meaning encoding and decoding can be performed more quickly, reducing the latency. The optimized JPEG2000 codec, such as used in AVPro MXNet Revolution II, supports Ultra-Low Latency (ULL) mode, achieving latency as low as 2-4 ms. If customers are unable to adopt a 10G network for their AV over IP solutions due to cost concerns, while latency remains a critical consideration, the AVPro MXNet Revolution II is the most cost-effective choice. To enable ULL mode, please contact AVPro's technical department for assistance.
- Lightweight Compression or No Compression: Solutions like SDVoE (Software-Defined Video over Ethernet) use very lightweight compression, achieving a compression ratio close to 1.4:1 to 1.3:1 for inputs with 4K60Hz resolution, resulting in a latency of approximately 100 microseconds (μs) in Genlock mode. For inputs with 4K30Hz resolution or below, SDVoE utilizes uncompressed transmission, further reducing latency. Based on our testing, the latency is virtually undetectable, which is why SDVoE is often described as delivering near-zero latency.
The compression ratio directly impacts the latency because higher compression typically involves more computation during both encoding and decoding. This computational load directly translates to increased latency. Conversely, minimal compression means the codec can quickly process the video signal, resulting in faster transmission.
End-to-End System Latency
End-to-end system latency in AV over IP refers to the entire process: from the signal source, through the HDMI cable to the encoder, then over Ethernet to a network switch, through another Ethernet link to the decoder, and finally from the decoder through HDMI to the display. In such a system, every component contributes to the total latency:
- HDMI Cables: Typically add negligible latency, usually around 5-10 nanoseconds (ns) per meter. The overall latency contribution is minimal unless very long HDMI cables are used.
- Ethernet Cables: Also add minimal latency, generally measured in the range of 5-50 nanoseconds (ns) per meter, depending on cable quality and length.
- Network Switch: The latency introduced by a network switch depends on its internal forwarding and queuing mechanisms, often adding between 3-10 microseconds (μs) per switch, depending on the model and network traffic conditions. Managed switches may have slightly higher latency due to additional processing.
- Encoder/Decoder Processing: The biggest contributors to latency are the encoder and decoder processes, which depend on the codec used and the processing power available. For instance, H.264/265 encoding may add 30-50 ms of latency, while lightweight codecs like SDVoE add less than 0.1 ms.
- Display Device: The display itself also introduces latency, typically between 10-30 ms, depending on the processing involved in scaling and rendering the image. Some professional displays have lower latency modes to minimize this.
To get an accurate measure of latency, you need to consider all these components together, which results in what is referred to as end-to-end latency.
Testing Latency with Murideo SIXA and SIXG
To measure the end-to-end system latency, professional tools like the Murideo SIXA (analyzer) and SIXG (generator) can be used. The Murideo SIXG acts as the signal source, while the SIXA serves as the sink device at the display end. The test setup involves the following equipment:
Encoder: AC-MXNET-10G-TCVR
Decoder: AC-MXNET-10G-TCVR
Network Switch: AC-MXNET-10G-SW24C
The test process is as follows:
- Connect the SIX-G to an INPUT of the SDVoE encoder (AC-MXNET-10G-TCVR) with an HDMI cable.
- Connect the SIX-A to the OUTPUT of the SDVoE decoder with an HDMI cable.
- Using a 3.5mm analog stereo cable, connect the SIX-G to the SIX-A using the SYNC port on both devices.
- Navigate to the SYNC & LATENCY TEST menu.
- Press the OK button.
- Read the results.
Test Result: When the signal source SIX-G is set to 3840x2160@30Hz in Passthrough mode, the entire end-to-end system latency is approximately 0.1 ms (100 microseconds). When the signal source SIX-G is set to 3840x2160@60Hz in Passthrough mode, the entire system's end-to-end latency is less than 0.1 ms. The final reading on the SIX-A is 0.0 ms.
Conclusion: Based on the end-to-end test results of the 10G SDVoE system, the overall system latency is extremely low, making it ideal for latency-sensitive applications. It is important to note that the display device is often the main contributor to system latency, so selecting a low-latency monitor or TV is crucial. The rest of the system—including the encoder, network switch, and decoder—adds negligible delay.
Latency Requirements Across Different Vertical Industries:
Different industries have varying requirements when it comes to latency:
- Broadcast and Live Events: These environments require Zero Latency (≤100 μs) since any visible delay would be unacceptable. SDVoE is well-suited for these applications.
- Medical and Surgical Applications: Also demand Zero Latency due to the need for precision and real-time feedback during surgeries.
- Corporate and Education: Often tolerate low latency (sub-50 ms), as delays of a single frame are not noticeable in presentations or classroom settings.
- Digital Signage: Typically tolerates higher latency, making high compression codecs like H.264/265 viable options, as cost-efficiency is prioritized over near-instantaneous display.
- Esports and Gaming Arenas: Require Zero Latency to ensure smooth and responsive gameplay experiences. SDVoE's low latency performance is ideal for such environments where even minor delays can impact player experience.
- Simulation and Training Systems: High-performance simulators, such as those used in aviation or military training, demand Zero Latency to create realistic and responsive simulations. SDVoE's capabilities are well-suited for these scenarios where accuracy and minimal delay are critical.
- Command and Control Centers: In environments where operators need to make real-time decisions, such as security monitoring or emergency response centers, Zero Latency is crucial. SDVoE's near-zero latency helps ensure that operators can respond quickly based on real-time video feeds.
Related Websites AVPro Edge • Murideo • SDVoE - Notes - 1 millisecond (ms) = 1,000 microseconds (μs) • 1 microsecond (μs) = 1,000 nanoseconds (ns)
Download the White Paper for a detailed resource you can reference anytime.
About AVPro
AVPro Global is a privately held American company headquartered in Sioux Falls, SD, with satellite offices
in St. Petersburg, FL, Seattle, WA, a European office in Bratislava, Slovakia, and an office in Shenzhen, China.
The five AVPro Global audio/video-centric brands - AVPro Edge, AudioControl Pro, Bullet Train, Murideo, and
Thenaudio – combine their respective strengths to create the fastest-growing AV signal distribution company
in the residential and commercial systems integration spaces, perfectly complementing AVPro Global Founder
and CEO Jeff Murray’s vision, “To be recognized as the best AV signal distribution and control products and
services company on the planet.”
Kevin Wang
VP, MXnet Systems Group
A seasoned expert in professional audio/
video, IT, and network technologies,
Kevin Wang joined AVPro Global in 2020
as VP of Product Development, where
he catapulted the MXnet ecosystem to
a prominent AV over IP contender in the
worldwide market. Previously, Wang was
a product manager and a COO, and the
co-founder of C2SEC, an AI-based cyber
risk assessment SaaS platform. He has a
bachelor’s degree from Tianjin University.
