Editorial

Special issue on deep learning in distributed and networked complex systems

Networking technologies have been widely used for data communication in complex dynamical systems due to a low-cost and efficient solution for information processing, which generates new configurations of industrial systems and leads to the emergence of distributed and networked complex systems (DNCSs). Such a DNCS has applications in a broad range of areas, such as wireless sensor networks, multi-area power systems, mobile robotics, intelligent transportation systems and so on. However, the implementation and design of practical DNCSs pose several challenges related to real-time, reliable and safe sensing, fusion, processing and communication of the large volumes of data due to large scale, spatial deployment, increased interconnectivity, network resource constraints and potential cyber-physical attacks in DNCSs. As a consequence, traditional information processing tools may be impractical or fail to tackle several analysis and synthesis problems of DNCSs because their performance does not improve when dealing with the challenging issues above. In contrast, embedding deep learning into the information processing for DNCSs promises numerous benefits. For example, information can be distilled more efficiently, increasingly abstract correlations can be obtained from the data without excessive pre-processing effort, spatially deployed subsystems can be monitored more reliably and safely, and algorithms can be implemented more accurately. There is no doubt that deep learning facilitates the performance analysis and synthesis of DNCSs with high accuracy, enhanced reliability, safety and resilience, overcoming the run-time limitations of traditional information processing tools. Therefore, it is profound to understand how to effectively and efficiently apply deep learning techniques with networking technologies to modern DNCSs with a large number of distributed sensors, controllers and actuators, which renders several fundamental problems regarding real-time and intelligent information processing significant.