With the development of more advanced and efficient control algorithms and communication architectures, UAVs and networks thereof (swarms) now find applications in nearly all possible environments and scenarios. There exist numerous schemes which accommodate routing for such networks, many of which are specifically designed for distinct use-cases. Validation and evaluation of routing schemes is implemented for the most part using simulation software. This approach is however incapable of considering real-life noise, radio propagation models, channel bit error rate and signal-to-noise ratio. Most importantly, existing frameworks or simulation software cannot sense physical-layer related information regarding power consumption which an increasing number of routing protocols utilize as a metric. The work presented in this paper contributes to the analysis of already existing routing scheme evaluation frameworks and testbeds and proposes an efficient, universal and standardized hardware testbed. Additionally, three interface modes aimed at evaluation under different scenarios are provided.

Advances in the fields of networking, broadband communications and demand for high-fidelity low-latency last-mile communications have rendered as-efficient-as-possible relaying methods more necessary than ever. This paper investigates the possibility of the utilization of cellular-enabled drones as aerial base stations in next-generation cellular networks. Flying ad hoc networks (FANETs) acting as clusters of deployable relays for the on-demand extension of broadband connectivity constitute a promising scenario in the domain of next-generation high-availability communications. Matters of mobility, handover efficiency, energy availability, optimal positioning and node localization as well as respective multi-objective optimizations are discussed in detail, with their core ideas defining the structure of the work at hand. This paper examines improvements to the existing cellular network core to support novel use-cases and lower the operation costs of diverse ad hoc deployments.

No abstract

Developments in the field of neural networks, deep learning, and increases in computing systems’ capacity have allowed for a significant performance boost in scene semantic information extraction algorithms and their respective mechanisms. The work presented in this paper investigates the performance of various object classification- recognition frameworks and proposes a novel framework, which incorporates Super-Resolution as a preprocessing method, along with YOLO/Retina as the deep neural network component. The resulting scene analysis framework was fine-tuned and benchmarked using the COCO dataset, with the results being encouraging. The presented framework can potentially be utilized, not only in still image recognition scenarios but also in video processing.