Ultra-low-latency video over 60 GHz wireless system

Toward the realisation of Society 5.0 and Industry 4.0, we have been seeing growing demand in the recent years for video applications that require real-time operation, such as hazard detection and remote control at various workplaces and factories. In particular, real-time wireless video transmission devices, which allow flexibility in installation and incorporation into mobile devices, are in great demand. 

In this experiment, an endpoint AI device being developed by TeraPixel Technologies connected with low-latency JPEG XS IP developed by intoPIX, was combined with 60 GHz millimeter-wave wireless communication module developed by Fujikura to transmit a video in real indoor and outdoor environments.

The results show that a high-quality (Full HD: 1920×1080 at 60 fps), ultra-low-latency video has been successfully transmitted. This has been enabled by combining the following features of the JPEG XS coding built in the endpoint AI device and the 60 GHz millimeter-wave wireless communication module.

Features of JPEG XS coding built in the endpoint AI device

• Low-latency compression/decompression (less than 0.5 ms) 

• High-quality visually lossless image with lightweight HD/4K/8K codec IP: TicoXS 

Features of 60 GHz millimeter-wave wireless communication module

• High-speed communication (> 3 Gbps) equal to that with optical fibers 

• Low transmission latency (around 2 ms) 

Furthermore, an experiment for video transmission from a camera mounted on a moving vehicle was conducted using the automatic beamforming and long-distance (> 500m) transmission performance of Fujikura's 60 GHz millimeter-wave wireless communication module.

Aiming at enterprise customers who pursue smartification in various workplaces such as factories, the three companies will jointly proceed with developing cutting-edge millimeter-wave video transmission products and services. The key is combining the license-free 60 GHz band communication system, which enables low-cost, high-speed network construction, and the endpoint AI equipment, which allows low-latency, high-quality image processing and AI-based image recognition.