Dynamic Placement of Mobile RSUs for Auction-Based Task Offloading in Urban Vehicular Clouds: A Smart Mobility Framework
Abstract/Description/Artist Statement
Smart mobility increasingly relies on real-time analytics, cooperative perception, and latency-sensitive computation near the network edge. Vehicular cloud computing is attractive because it can exploit the unused computational resources of vehicles dispersed across a city. However, effective utilization is challenged by mobility-driven churn, intermittent wireless links, and the lack of coordination infrastructure that can reliably support bidding, input delivery, monitoring, and output collection at scale. This paper proposes an integrated framework that couples (i) rolling-horizon placement of mobile roadside units (mRSUs) with (ii) a truthful, reliability-aware reverse auction for vehicular task offloading. The city contains many stationary RSUs that provide baseline coverage, while a smaller fleet of mRSUs dynamically repositions to demand hotspots and coverage gaps. We formulate mRSU placement as a rolling-horizon mixed-integer optimization that maximizes incremental coordination coverage while minimizing mobility cost and fairness penalties, with redundancy constraints for resilience. On top of this hybrid RSU layer, a multi-attribute sealed-bid reverse auction allocates stochastic tasks to heterogeneous vehicles while accounting for readiness, departure uncertainty, compute reliability, and link success probability. The auction mechanism is shown to satisfy individual rationality, incentive compatibility under fixed admissibility, and budget balance, while deliberately trading allocation efficiency for deadline-robust task completion. A preemptive migration and re-auction policy further mitigates failures due to mobility and link degradation. Simulation results demonstrate improved coverage efficiency, reduced fairness penalty accumulation, and stronger coordination robustness compared to static and reactive baselines.
Faculty Advisor/Mentor
Stephan Olariu
Faculty Advisor/Mentor Email
solariu@odu.edu
Faculty Advisor/Mentor Department
Computer Science
College/School Affiliation
College of Sciences
Student Level Group
Graduate/Professional
Presentation Type
Oral Presentation
Dynamic Placement of Mobile RSUs for Auction-Based Task Offloading in Urban Vehicular Clouds: A Smart Mobility Framework
Smart mobility increasingly relies on real-time analytics, cooperative perception, and latency-sensitive computation near the network edge. Vehicular cloud computing is attractive because it can exploit the unused computational resources of vehicles dispersed across a city. However, effective utilization is challenged by mobility-driven churn, intermittent wireless links, and the lack of coordination infrastructure that can reliably support bidding, input delivery, monitoring, and output collection at scale. This paper proposes an integrated framework that couples (i) rolling-horizon placement of mobile roadside units (mRSUs) with (ii) a truthful, reliability-aware reverse auction for vehicular task offloading. The city contains many stationary RSUs that provide baseline coverage, while a smaller fleet of mRSUs dynamically repositions to demand hotspots and coverage gaps. We formulate mRSU placement as a rolling-horizon mixed-integer optimization that maximizes incremental coordination coverage while minimizing mobility cost and fairness penalties, with redundancy constraints for resilience. On top of this hybrid RSU layer, a multi-attribute sealed-bid reverse auction allocates stochastic tasks to heterogeneous vehicles while accounting for readiness, departure uncertainty, compute reliability, and link success probability. The auction mechanism is shown to satisfy individual rationality, incentive compatibility under fixed admissibility, and budget balance, while deliberately trading allocation efficiency for deadline-robust task completion. A preemptive migration and re-auction policy further mitigates failures due to mobility and link degradation. Simulation results demonstrate improved coverage efficiency, reduced fairness penalty accumulation, and stronger coordination robustness compared to static and reactive baselines.