Optimal Link Resource Allocation and Pricing in Computility Networks

Abstract

Computility networks have emerged as a critical infrastructure for large-scale, data-intensive task execution, where computing and network resources must be jointly allocated across multiple providers and network operators. Although prior studies have mainly focused on computing resource allocation, the allocation and pricing of network resources have received relatively limited attention, due to the topology-dependent and flow-based nature of network resources. To tackle this issue, we investigate the problem of network link allocation and pricing in Computility networks, with a focus on mechanism design. We first formulate a mixed-integer optimization model that minimizes users' network transmission costs under joint network-flow and computing-resource constraints. Based on this model, we propose VCG-LAPM(VCG-based link allocation and pricing mechanism). VCG-LAPM employs a successive shortest-path min-cost flow algorithm to compute efficient link allocations and adopts a reverse VCG pricing rule to ensure incentive compatibility and individual rationality in theory. To improve robustness in the presence of critical links, a price-cap rule is further introduced. Experimental results show that VCG-LAPM can achieve nearoptimal user payments and significantly improves task acceptance rates compared with single-operator constrained schemes across varying network scales and requirement intensities.