Security-aware computation offloading in internet of vehicles: a multi-agent reinforcement learning algorithm with attention mechanism
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
Limited by the finite computation resources in Internet of Vehicles (IoV), it is necessary to facilitate the capacity of vehicles using computation offloading. The traditional approaches focus on maximizing utilities of computation offloading while fail to take into consideration the security of offloading strategy, which cannot be used for IoV environments directly. In addition, unsafe strategies can cause interruptions in task offloading and waste computation resources. In view of this, this paper ensures secure task offloading by mitigating interruptions caused by communication constraints such as limited transmission distance. To this end, we propose a two-layer security-aware computation offloading framework and design a multi-agent proximal policy optimization algorithm with attention mechanism (named MAPPOAM) to address the aforementioned issue. More specifically, in the feature extraction layer, the attention mechanism is utilized to obtain the relationship between the communication time and movement data of the vehicles and extract the feature of data. In the security decision layer, the agents output the offloading strategy based on the observation information. Subsequently, the algorithm in the security constraint layer analyzes the relationship between the task execution time constraints and the extracted feature to evaluate the safety of this strategy. Eventually, the agent outputs a strategy that ensures secure completion of the offloading process. The experiments with the real dataset show that MAPPOAM can effectively ensure the security of decisions compare with other baseline methods and provide valuable practical applications. Specifically, MAPPOAM outperforms the baseline algorithms, reducing both latency and energy consumption by 17.5% to 63.1%. Additionally, the security decision layer contributes to a 60.4% reduction in offloading interruptions.
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 | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 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 | 0.000 |
| 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