In the field of smart communication, NEC is engaged in initiatives to improve the quality of control communication (remote control) as well as video and audio communication (multimedia communication) delivered via the Internet and mobile networks.
Best-effort networks such as Internet and mobile networks affect service quality because communications throughput varies significantly depending on such factors as radio wave conditions and usage by other users. For this reason, users of the Internet-based video streaming services and cloud services widely available today are facing issues of deteriorated service quality such as communications delays and low image quality.
NEC Central Research Laboratories is conducting research into communications optimization technologies to enable efficient, optimal communications even via Internet and wireless networks, which are subject to dramatic fluctuations in quality. We have now realized high-accuracy future prediction for Internet communication which was difficult to achieve up to now. A variety of high quality cloud services can be realized by controlling services according to predictions of communications conditions.
High-accuracy and real-time remote control of robots (Adaptive Remote Control)
As the world migrates towards IoT (Internet of Things), robots are being controlled remotely via communication networks. Use of wireless IP (Internet Protocol) communications is becoming particularly prevalent in view of connectivity and mobility. However, using wireless IP communications to remotely control robots that require high-accuracy real-time control is subject to communication delays which can prevent sensor information and control commands from arriving in a timely manner. This in turn has hindered the accurate control of robots.
In response to this situation, NEC developed "Adaptive Remote Control," which conducts highly accurate predictions of communication delays stemming from wireless IP communications, and achieves high-accuracy and real-time remote control of robots by estimating the future state of the robots on the basis of the delay prediction.
Adaptive Remote Control has the following two technical characteristics.
(1) Highly accurate predictions of communication delays
An analogous relationship was discovered between the mechanism that causes delays in a communication network and the mechanism for transition of molecule states. Applying the analytical method used on the transition of molecule states has enabled highly accurate predictions of communication delays.
- (*1)Yasuda, Yoshida: "Prediction of round-trip delay based on the two-state model," IN Study Group, The Institute of Electronics, Information and Communication Engineers, Sept. 2017
(2) Proactive control based on the predictions of communication delays
The above communication delay prediction technology enables predictions regarding the amount of time that has passed to receive feedback from the robots, and the length of delay for the control commands to reach the robots. By using the predictions to accurately estimate the state of the robot at the time the control commands arrive then using the estimates to implement control ahead of the delays, it becomes possible to achieve accurate real-time control even from remote locations where communication delays occur.(*2,3)
- (*2)H. Yoshida, K. Kumagai, and K. Satoda, "Dynamic State-Predictive Control for a Remote Control System with Large Delay Fluctuation," IEEE ICCE, Las Vegas, Jan. 2018.
- (*3)K. Kumagai, H. Yoshida, and K. Satoda, "Adaptive Remote Control of a Mobile Robot System with Delay Fluctuation," IEEE ICCE, Las Vegas, Jan. 2018.
Ref.: Collaborative research with Nidec
Video distribution technology to support surveillance, emergency services and disaster prevention and response (adaptive video feed control)
With conventional surveillance that depends on fixed surveillance cameras and the eyes of security guards, information sharing between surveillance headquarters and the location being monitored has tended to be insufficient, and the risk of human oversight or blind spots is high. Meanwhile, use of video distribution, while finding some traction in high-end surveillance systems, is underutilized in busy communications environments such as event locations because it is not possible to distribute video of the quality required for surveillance in real time. NEC has developed "adaptive video feed control technology" capable of distributing video of sufficient quality to assess the situation at the location being monitored in real time, by predicting communications throughput variations up to 10 seconds in advance and adaptively controlling the number of frames and the video compression rate. This technology can significantly contribute to improving the safety and security of society, not only for surveillance, but also by enabling accurate relaying of patient conditions in medical emergencies and information about conditions in disaster-stricken areas.
(1) Communications throughput (band) prediction
AI analysis of communications throughput variation over the past one minute. Constructs a prediction model in real time and makes a prediction of future communications throughput variation from proportional spread.
(2) Adaptive rate control
Controls compression rate and number of frames based on the communications throughput prediction. Avoids video quality degradation due to excessive compression to realize usable quality video distribution for surveillance, emergency medical and disaster relief services.