Joint Data Caching and Computation Offloading in UAV-Assisted Internet of Vehicles via Federated Deep Reinforcement Learning
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Due to the dense buildings around the macro base stations (MBSes) and the hotspot requests within particular area (e.g., traffic intersections), it is a challenging task for Quality of Service (QoS) guarantee in Internet of Vehicle (IoV). To address these challenges, unmanned aerial vehicles (UAVs) can be integrated into mobile edge computing (MEC) for IoV by leveraging their advantages of mobile flexibility, low price, and line-of-sight (LoS) communication links. In this paper, we establish a joint UAV-assisted IoV scenario, where both UAVs and MBSes can provide computation and data caching services for smart vehicles. Then, we formulate a joint optimization problem for dynamic data caching and computation offloading, aiming to minimize the average task processing delay and maximize the UAV cache hit ratio. By applying deep reinforcement learning (DRL) techniques, we design an intelligent data caching and computation offloading (IDCCO) algorithm to deal with large-scale and continuous state and action spaces. Furthermore, in order to accelerate the convergence speed of DRL model training while protecting the privacy of original user data in IoV, we propose a distributed training mechanism based on Federated Learning (FL), where the DRL model training is performed locally on UAV and global parameter aggregation is performed on MBS. Finally, extensive experiments are conducted, and the experimental results demonstrate the superiority of our approach over several comparative algorithms in shortening the training time, reducing the task processing delay, and maximizing the cache hit ratio.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it