Periodic Reporting for period 1 - GAMODRONE (Game of Drones: Multimedia Streaming with Flying Terminals in Next Generation Mobile Networks)
Okres sprawozdawczy: 2017-07-01 do 2019-06-30
Recent years have witnessed an unprecedented growth in the volume of Internet data flow as well as in the number of devices connected to the Internet. Apart from traditional handheld mobile terminals, the use of civilian unmanned aerial vehicles (UAV) that allow capturing and streaming video in real-time is steadily increasing. The technology for civil drones is maturing and there is potential for significant growth and job creation. On some estimates in the next 10 years it could be worth around 10% of the whole aviation market. Such rapidly growing market is going to pose important challenges on the future 5G/6G terrestrial networks. Although for the time being civilian UAV platforms are mainly meant to stream first person view (FPV) videos directly to the user terminal on unlicensed frequency bands, streaming directly on the Internet or private networks is going to become a concrete possibility soon. If on the one hand drones constitute a challengefrom a communication perspective, on the other hand they can be a potential asset for next generation mobile networks, by helping to provide missing coverage or “on demand” additional capacity.
The goal of the present project was two-fold. The first goal was to develop new approaches to jointly optimize video and image production and transmission from civil UAV’s over wireless networks with focus of UAV-specific requirements such as low delay requirements and high mobility. The second goal was to exploit the unique 3D mobility capability of drones to assist mobile networks by providing additional capacity and missing coverage.
Both goals of the project have been achieved by delivering original results, ranging from fundamental theoretical limits to video compression and streaming to practical low-complexity algorithms and development of proof-of-concept prototypes. Several top-ranked publications have been published and others are being finalized for submission.
The goal of the present project was two-fold. The first goal was to develop new approaches to jointly optimize video and image production and transmission from civil UAV’s over wireless networks with focus of UAV-specific requirements such as low delay requirements and high mobility. The second goal was to exploit the unique 3D mobility capability of drones to assist mobile networks by providing additional capacity and missing coverage.
Both goals of the project have been achieved by delivering original results, ranging from fundamental theoretical limits to video compression and streaming to practical low-complexity algorithms and development of proof-of-concept prototypes. Several top-ranked publications have been published and others are being finalized for submission.
With respect to the first goal, we started by deriving a mathematical model capable of capturing the main characteristics of videos shot from high-mobility platforms such as drones. Then we moved to the study of original solutions for streaming of videos from UAV’s under stringent delay and video quality constraints using tools from information theory. Based on our theoretical results, we proposed a low-complexity solution to bring such results closer to a practical implementation.
As of the second goal, we contributed to one of the subjects at the forefront of research in next generation mobile networks which is caching. Caching consists in storing some (typically the most popular) content at the edge of the communication network (in our case onboard the drones), rather than in the core network, in order to mitigate network congestion during peak hours. We focused on the fundamental problem of guaranteeing a given quality of service in the presence of interference and developed an optimal cache allocation algorithm as well as an efficient low-complexity solution. Another topic related to the second goal is the deployment of drones in mobile networks to improve nodes localization for the Internet of Things (IoT). Such topic has been studied in collaboration with Swisscom and the LIS robotics laboratory of EPFL and lead to the development of a proof-of-concept system which showed good results on field tests.
As of the second goal, we contributed to one of the subjects at the forefront of research in next generation mobile networks which is caching. Caching consists in storing some (typically the most popular) content at the edge of the communication network (in our case onboard the drones), rather than in the core network, in order to mitigate network congestion during peak hours. We focused on the fundamental problem of guaranteeing a given quality of service in the presence of interference and developed an optimal cache allocation algorithm as well as an efficient low-complexity solution. Another topic related to the second goal is the deployment of drones in mobile networks to improve nodes localization for the Internet of Things (IoT). Such topic has been studied in collaboration with Swisscom and the LIS robotics laboratory of EPFL and lead to the development of a proof-of-concept system which showed good results on field tests.
Original contributions in the frame of both project’s goals have been produced. Specifically, we proposed an original model for videos captured from UAV’s and studied the problem of streaming from UAV’s with real-time and playback distortion requirements both from a theoretical and an implementation perspective, proposing an effective original solution to tackle the problem. We also studied the problem of optimal cache allocation in UAV-assisted mobile networks affected by interference where different per-file rate constraints are in place. Our original results will contribute to the development of streaming technologies that are specific for UAV’s as well as to the optimization of cache placement in caching-enabled UAV’s for next generation mobile networks.