[Paper] Are Bus-Mounted Edge Servers Feasible?

Source: arXiv - 2512.05543v1

Overview

Edge computing is becoming a cornerstone for the Internet of Vehicles (IoV), but static edge nodes (e.g., RSUs or base stations) can’t keep up with the highly dynamic spatial‑temporal demand of moving vehicles. This paper investigates whether mounting edge servers on city buses can provide a flexible, cost‑effective complement to fixed infrastructure. Using real‑world mobility and traffic data from Shanghai, the authors demonstrate that a modest fleet of bus‑mounted servers can dramatically improve service coverage and adapt to fluctuating demand.

Key Contributions

• Empirical coverage analysis using large‑scale Shanghai bus, taxi, and telecom datasets, showing that buses traverse a large fraction of the city’s demand hotspots.
• Mathematical formulation of the bus‑selection problem: maximize the number of demand points covered given a limited budget (i.e., a limited number of buses to equip).
• Greedy heuristic algorithm that efficiently selects the most “useful” buses, achieving near‑optimal coverage with low computational overhead.
• Trace‑driven simulation that incorporates realistic constraints (server capacity, purchase quantity) and validates the algorithm’s ability to handle dynamic vehicular demand.
• Feasibility claim that bus‑mounted edge servers are a practical, beneficial addition to urban vehicular networks.

Methodology

1. Data Collection & Pre‑processing

   • Mobility traces: GPS logs of Shanghai’s public buses and taxis.
   • Demand traces: Telecom records indicating where and when mobile devices (proxy for vehicles) generate traffic.
   • The city is discretized into small geographic cells; each cell records the number of demand events over time.

2. Coverage Modeling

   • A bus is considered to cover a cell if it passes within a predefined radio range (e.g., 300 m) during a time slot.
   • Coverage is time‑dependent: a bus may serve a cell only when it is physically present.

3. Optimization Problem

   • Objective: Choose up to K buses (budget) to maximize the total number of demand events covered across all cells and time slots.
   • Constraints: Each selected bus has a finite compute capacity; demand exceeding this capacity is considered uncovered.

4. Greedy Heuristic

   • Iteratively pick the bus that adds the largest marginal increase in covered demand (taking capacity into account).
   • Stop when K buses are selected or no further improvement is possible.

5. Simulation Framework

   • Replay the real traces, assign the selected buses as edge servers, and measure coverage, server load, and latency proxies.
   • Compare against baselines: (i) only fixed RSUs, (ii) random bus selection, and (iii) an optimal (exhaustive) solution on a small subset.

Results & Findings

• Coverage boost: The greedy algorithm captures roughly 30 % more demand than static RSUs alone, confirming that buses can fill geographic and temporal gaps.
• Elastic capacity: Because buses move, the same physical server can serve multiple distant hotspots at different times, reducing the need for dense RSU deployment.
• Scalability: Even with a modest budget (e.g., equipping 30 out of ~1,200 buses), the system achieves near‑optimal coverage; adding more buses yields diminishing returns, indicating a sweet spot for investment.

Practical Implications

• Cost‑Effective Edge Expansion: Municipalities or telecom operators can augment existing edge infrastructure by retrofitting a limited number of buses, avoiding costly construction of new RSU sites.
• Dynamic Service Placement: Developers can design IoV applications (e.g., real‑time traffic analytics, AR navigation, V2X safety alerts) that offload compute to the nearest bus‑mounted server, improving latency during peak traffic periods.
• Resource Orchestration: Cloud‑edge orchestration platforms can treat bus‑mounted servers as “mobile nodes,” scheduling workloads based on predicted bus routes and capacity, similar to edge‑as‑a‑service.
• Policy & Planning: Urban planners can incorporate bus routes into edge‑network design tools, optimizing route planning not just for passenger flow but also for computational coverage.

Limitations & Future Work

• Route Predictability: The study assumes relatively stable bus schedules; unexpected detours or traffic jams could degrade coverage.
• Hardware Constraints: Real‑world deployment must address power, cooling, and ruggedization of edge hardware on moving vehicles.
• Security & Mobility Management: Frequent handovers and the mobile nature of the servers raise challenges for authentication, data privacy, and seamless service continuity.
• Scalability to Larger Cities: While Shanghai provides a rich dataset, the model needs validation in cities with different transit topologies (e.g., sparse bus networks, mixed‑mode transport).

Future research directions include adaptive routing algorithms that dynamically reassign workloads based on real‑time bus telemetry, integration with other mobile platforms (e.g., delivery drones, ride‑hailing cars), and prototype deployments to evaluate energy consumption and maintenance overhead.

Authors

• Xuezhi Li
• Jiancong He
• Ming Xie
• Xuyang Chen
• Le Chang
• Li Jiang
• Gui Gui

Paper Information

• arXiv ID: 2512.05543v1
• Categories: cs.DC
• Published: December 5, 2025
• PDF: Download PDF