Call for Papers (Special Sessions)

The need for higher data-rates and more ubiquitous connectivity for an ever-increasing number of wirelessly connected devices motivates the exploration of uncharted spectral bands. In this context, terahertz (THz)-band (broadly, from 100 GHz to 10 THz) communication is envisioned as a key wireless technology of the next decade. Nevertheless, there are several roadblocks that need to be overcome to tap in the THz band, ranging from the lack of high-power THz sources, high sensitivity detectors and steerable directional antenna systems, to advanced signal processing, communication and networking techniques that can make the most of the ultra-broadband THz channel while overcoming the challenging propagation characteristics of THz waves. The talks in this session will provide a peak at the state of the art in multiple inter-disciplinary problems that need to be addressed to conquer the spectrum above 100 GHz.

Radio frequency (RF) machine learning (ML) has been gaining momentum, showing promise in numerous applications including modulation recognition and RF fingerprinting.  Most research, however, has focused on initial proof of concept studies, showing promise for a single application considering only a few variables, i.e., signal to noise ratio, population size, etc.  This session looks at research that goes beyond these first order proofs of concept.  This includes operational challenges, such as the presence of signal interference, varying channel conditions, lack of realistic training data, or algorithm portability across hardware.  This also includes incorporating RF ML application(s) into a more complete system with focus on understanding cascading performance effects, real-time constraints, and system degradation from novel inputs, data drift, or adversarial techniques.  The talks of this session will create a dialog for how to make RF ML solutions more efficient or robust.

With the enhancements in 5G and upcoming 5G+ technologies, intelligent ML driven network management, automation and control of Cyber-Physical Systems have attained a big importance in order to serve for more reliable and cost effective applications. Here, 5G+ network control automation is the process of making the cyber-physical system to have self-organized properties regarding the challenging management functions. However, traditional network automation methods can not afford the desired performance levels. This is because of the fact that traditional network automation approaches lack capability in understanding and interpreting the data. For example, in ultra dense 5G networks, end-to-end delay and the heterogeneity in IoT traffic load are the significant performance parameters to maintain a qualified network service. In such dense networks, non-uniform loading and unpredictable increase in the traffic flows draw attention to the need for intelligent load balancing strategies. Likewise, core delay and edge delay arise as the key performance metrics for wireless dense networks. Although the main target in these networks is to improve the service quality and end user experience, the space and time constraints in resources should also be observed by an intelligent agent. Furthermore, in multi-connectivity scenarios for 5G+ based machine-to-machine or IoT enabled cyber-physical system network application, the mobility handling problem arises as the primary performance indicator. Here as well, intelligent mobility management schemes with forecasting; the device or vehicle connectivity status are required. Despite the network automation function being applied in all these 5G+ infrastructured network applications, there is a lack of learning ability for this function leading to significant performance degradation.  For this reason, to meet the goal of ‘intelligent and self-driven networks’, we need fine grained intelligence aware network automation approaches. In this way, we can handle such multi-faceted network management problems for 5G+ infrastructured environments. With these motivations in mind, in this special session, we investigate the need of fine grained intelligence aware network approaches, both in edge and core perspectives. These include but not limited to 5G+ ad-hoc and centralized cyber-physical network control automation challenges such as software defined and AI driven topology management, coverage areas issues, protocol design, fault diagnosis, end-to-end performance monitoring as well as network security.