As next-generation applications like autonomous driving, real-time navigation, and mobile biomedical telemetry become ubiquitous, they demand unprecedented computational power and ultra-low latency. Traditional mobile edge computing relies on fixed infrastructure, such as Roadside Units (RSUs) and base stations, to handle these tasks. However, as noted by Guoling Liang et al. and Liang Liu et al., fixed edge servers frequently suffer from localized congestion and high latency during peak traffic hours due to their rigid physical constraints and high deployment costs.
To solve this bottleneck, a powerful research direction has emerged: opportunistic resource pooling. Instead of treating the influx of mobile devices and vehicles as a source of network congestion, this paradigm flips the script by treating traffic density as a distributed, elastic cloud of computational infrastructure.
The Core Intuition: Traffic as an Asset
Modern electric and autonomous vehicles possess massive onboard computing power and battery capacity that remain completely idle during long stretches of travel, charging, or parking (Jiming Chen et al.). Traditional offloading models treat these vehicles merely as clients, ignoring this latent capacity.
Opportunistic resource pooling aggregates these underutilized resources into a shared, decentralized infrastructure layer. As Salman Raza et al. demonstrate, increasing the density of participating mobile nodes actually decreases total system costs. High-traffic environments are effectively transformed into a distributed moving cloud. This concept, often formalized as "Vehicle as a Service" (VaaS) or "Fog Computing on Wheels," allows cities to harvest unutilized local resources at the far edge, bypassing the need for expensive fixed 5G+ infrastructure deployments (Yiqin Deng et al.; Tom H. Luan et al.; Christophoros Christophorou et al.).
[Overloaded Fixed RSU] ──(Spillover Tasks)──> [Dynamic Vehicular Cloud]
│
┌─────────────────────────────┴─────────────────────────────┐
▼ ▼
[Parked Vehicles Pool] [Moving Vehicles Pool]
(Stable Far-Edge Micro-Datacenters) (Transient P2P/V2V Helpers)
The Dual-Engine Architecture
To make a volatile pool of moving parts behave like a reliable data center, the architecture splits resources into two distinct categories:
1. Parked Vehicles as Far-Edge Micro-Datacenters
Vehicles remain stationary for the vast majority of their lifespan. Research by Muhammad Waqas et al. and Jun Zhang et al. highlights that parked vehicles provide a highly stable, high-capacity resource pool compared to moving traffic. Because they face fewer power and mobility constraints, these stationary assets can be seamlessly integrated into standard multi-access edge computing (MEC) frameworks. Angelo Feraudo demonstrates that utilizing parked vehicles as far-edge micro-datacenters drastically reduces backbone network stress and extends edge services to the extreme network periphery without capital expenditure.
2. Moving Vehicles as Dynamic Coalitions
For real-time, localized tasks, nearby moving vehicles can dynamically cluster to pool their unused processing power (Yuguang Fang et al.). Managed by a local access point, these mobile nodes form transient coalitions where heavy computational tasks are distributed among peer devices via Vehicle-to-Vehicle (V2V) or Device-to-Device (D2D) communication (Renchao Xie et al.; Maria Gorlatova et al. et al.). This allows the network's total processing capacity to scale naturally with traffic density.
Why It Works: Joint Communication and Computation Cooperation
The mathematical and physical viability of this approach hinges on joint cooperation frameworks. Instead of treating communication and computation as isolated steps, modern systems optimize them simultaneously.
As pioneered by Xiaowen Cao et al., an idle device acts as a "helper" by both performing part of a user's computations and acting as a communication relay to offload the remainder to an access point. This dual-role capability bypasses communication dead zones and alleviates central congestion.
When an application generates intensive analytical tasks—such as processing real-time driver fatigue levels or biomedical telemetry (Honglong Chen et al.)—the task is partitioned. Parts of the task are executed locally, parts are sent to nearby moving helpers, and parts are routed to parked vehicle clusters.
By crowdsourcing this surplus battery and processing capacity, the system minimizes the energy burden on individual source vehicles and centralized infrastructure (Tai Manh Ho et al.). Empirical evaluations show that this cooperative strategy yields significant efficiency gains:
- Energy Savings: Cooperative vehicular computing achieves a 7% to 18% reduction in energy consumption compared to local processing (Junhua Wang et al.).
- Cost and Latency Reduction: Offloading tasks from overloaded edge nodes to nearby vehicular clusters can cut the combined cost of processing latency and energy consumption nearly in half (Rahul Yadav et al.).
Overcoming the Mobility Dilemma
The primary challenge to this paradigm is the "mobility dilemma"—the reality that fast-moving nodes create highly volatile topologies, risking task failures if a helper drives out of range mid-computation.
To resolve this, modern frameworks incorporate mobility-aware scheduling. For example, the scheduling algorithms designed by Salman Raza et al. explicitly calculate the "stay time" of moving cars to determine optimal task partitioning. Furthermore, by aggregating resources from multiple overlapping RSUs and parking structures into a single virtual resource pool, the network achieves elastic resource allocation (Honglong Chen et al.; Guoling Liang et al.). If one node moves away, neighboring nodes within the virtualized pool seamlessly absorb the load.
Ultimately, by shifting from rigid, centralized infrastructure to an opportunistic, crowdsourced model, this research direction provides a scalable, sustainable pathway toward blanket digital coverage in smart cities.
Key Foundational Work
Cost-Effective Resource Sharing in an Internet of Vehicles-Employed Mobile Edge Computing Environment (2018) — contributed 1 concept elements to this direction.
Guest editorial: fog computing on wheels (2018) — contributed 1 concept elements to this direction.
Joint Computation and Communication Cooperation for Mobile Edge Computing (2017) — contributed 1 concept elements to this direction.