.png){kind=icon}
.png){kind=icon}
Return
Fairness-efficiency tradeoffs in multiresource allocation for cloud–edge collaborative computing
.png){kind=icon}
PRE
AI
DOI：10.1016/j.future.2026.108694.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
In cloud–edge collaborative computing systems, heterogeneous servers, diverse multiresource demands, and bandwidth demand compression make it inherently difficult to balance fairness and efficiency in multiresource allocation. Traditional multiresource fairness allocation mechanisms pursue absolute fairness, leading to insufficient resource utilization and degraded system performance. To solve this problem, this paper proposes dominant resource soft fairness in cloud–edge collaborative computing (SDRF-CE) mechanism that achieves flexible and controllable fairness-efficiency tradeoffs in multiresource allocation for cloud–edge collaborative computing systems. SDRF-CE includes two sub-mechanisms: the soft fairness preallocation mechanism (SFPAM) and the maximizing overall efficiency mechanism (MOEM). SFPAM determines the minimum dominant share for each user, thereby ensuring a lower bound of fairness degree relative to an absolutely fair allocation. Subsequently, MOEM allocates resources with the goal of maximizing the total sum of dominant shares across all users, thereby improving allocation efficiency. Rigorous proofs show that SDRF-CE satisfies Pareto optimality and envy-freeness while also satisfying three relaxed properties: soft fairness, α -proportionality and β -strategy-proofness. Results of simulations driven by Google and Alibaba cluster traces show that SDRF-CE can improve allocation efficiency by up to 107% compared with strict fairness mechanisms while maintaining the expected fairness degree. The simulation results also show that the SDRF-CE algorithm has robustness and good operational efficiency, and can achieve effective and flexible tradeoffs between fairness and efficiency in allocation based on the expected fairness degree.
Journal
F
IF:
6.1
Papers:
6.8K
Citations:
2.3W