Introduction

For wireless microwave video transmission system, latency is a an important parameter. For developers, if the system delay can be reduced by 3ms, the cost may be increase by 10%. Today we will introduce you to a practical latency measure method.

First of all, let’s introduce two concepts:

• End to end latency: the whole time required from image captured from camera, wireless transmitted to receiving side and then displayed on monitor. As shown below:

  
  

• Camera Latency: Time required for the process that camera outputs video to monitor directly by HDMI cable connection. As shown below:

Measure the camera latency

• Prepare all the stuffs that you will need for this step:

• Measure steps:

• 1. Open the Only Stop Watch software and start timing

• 2. Connect Gopro camera to the HD display via a HDMI cable

• 3. Point the Gopro camera at the timer on computer screen. Simultaneously, you can get the timer on the display.

• 4. Adjust the position of the computer screen and monitor, so that you can capture two screen images at same time with mobile phone and take multiple photos in succession

The difference between two timers in the photo is the delay of GO Pro camera, which is recorded as X. As shown in FIG, go pro camera latency is 69ms. Based on many times calculating, the latency is about 65-120ms.

Secondly let’s measure the end to end latency

We need to repeat the above process  however the difference is Gopro camera need to be connected to the transmitter through HDMI, and the receiver will be connected to the monitor. Please refer to the figure below:

As follows end to end latency is 277ms (640-363)

At present, most of wireless high-definition digital image transmission system on the market have undergone encoding, modulation, transmission, demodulation, decoding, etc. Delays of different schemes will also be different.

The factor that has the greatest impact on latency is the video codec. Let’s take P60 (60 frames) image as an example. Time to encode and decode per frame is 16.7 ms. In order to ensure video smoothness, currently the popular encoding scheme is to first receive 1~3 frames for caching and then extract images from the cache and encode them. The decoding also buffers 1~3 frames and then outputs. So the codec time is the main part of the delay.

With the development of technology, some new technologies have emerged. Such as block coding, which can encode images while receiving. This way the cache delay can be saved. In addition, making use of FPGA parallel processing combined with block coding technology for image coding can much greater short the latency.