Fast and Fair Computation Offloading Management in a Swarm of Drones Using a Rating-Based Federated Learning Approach

Abstract

Today, unmanned aerial vehicles (UAVs) or drones are increasingly used to enable and support multi-access edge computing (MEC). However, transferring data between nodes in such dynamic networks implies considerable latency and energy consumption, which are significant issues for practical real-time applications. In this paper, we consider an autonomous swarm of heterogeneous drones. This is a general architecture that can be applied for applications that need in-field computation, e.g. real-time object detection in video streams. Collaborative computing in a swarm of drones has the potential to improve resource utilization in a real-time application i.e., each drone can execute computations locally or offload them to other drones. In such an approach, drones need to compete for using each other's resources; therefore, efficient orchestration of the communication and offloading at the swarm level is essential. The main problem investigated in this work is computation offloading between drones in a swarm. To tackle this problem, we propose a novel federated learning (FL)-based fast and fair offloading strategy with a rating method. The simulation results demonstrate the effectiveness of the proposed strategy over other existing methods and architectures with average improvements of -23% in energy consumption, -15% in latency, +18% in throughput, and +9% in fairness.INDEX TERMS Swarm of drones, Multi-access edge computing, Collaborative computing, Federated learning.

A Fairness-Enhanced Federated Learning Scheduling Mechanism for UAV-Assisted Emergency Communication

“…Therefore, we believe that the deviation of the measurement from the mean does not present a good picture of the fairness performance of the system. The Jain metric is more widely used and easier to calculate than the Gini coefficient [ 35 ]. It is calculated using the accuracy of the models on the private datasets, which to some extent reflects the extent to which the global model contributes to the local models.…”

Federated Reinforcement Learning for 6G Wireless Networks: Fundamentals, Challenges and Future Research Trends

“…Furthermore, in [200], the authors proposed a dualstage RL incentive system that inspires the UEs to train the global model utilizing huge amounts information sets to make the most of their rewards. Instead, in [201][202], the authors proposed an FL-based weather quality observing structure for safe UE crowdsensing. Then, a swarm of UAVs is utilized to estimate the weather quality, and the sensed information is utilized to train a lightweight cellular network model to forecast the exponent of weather quality.…”

Federated learning in cloud-edge collaborative architecture: key technologies, applications and challenges

“… ST ResNet n/a Secure and efficient Fantacci and Picano [ 97 ] Demand prediction Protect sensitive user data FedAvg MovieLens 1M & MovieLens 100K better than the approach based on chaos theory and deep learning. Taïk and Cherkaoui [ 98 ] Household load forecasting Protect user privacy FedAvg n/a significant gain in the network load Rahbari et al [ 99 ] UAV Improve resource utilization in real-time applications. Aggregate by scoring weight n/a better fairness & energy efficient Pham et al [ 100 ] UAV Improve the transmit power efficiency of UAVs Decomposition n/a Dramatically reduce drone launch power Chen et al [ 101 ] Augmented Reality Improve computational efficiency & Reduce latency CNN CIFAR-10 Fewer training iterations Hsu et al [ 102 ] Information Security Android malware detection SVM from NICT Outperforms centralized training systems.…”