Risk-Aware Scheduling and Placement for Bus-Stop Wireless Powered Communication Networks

초록

Scheduled electric buses can serve as mobile hybrid access points (HAPs) for bus-stop wireless powered communication networks (WPCNs), providing short energy-transfer and uplink opportunities to low-power passenger-carried devices waiting outside the bus. We consider a brief uplink before boarding; after the passenger boards, the corresponding device leaves the modeled bus-stop user set. Some devices may therefore end up with low terminal accumulated throughput, making lower-tail fairness a central design concern. Addressing this risk requires per-stop decisions: selecting users, choosing curbside bus placement, and allocating dwell time between downlink wireless power transfer and uplink information transmission. We formulate this as a stochastic multi-stage mixed-integer nonlinear program maximizing a Conditional Value-at-Risk (CVaR) objective over terminal accumulated user throughputs. To make the problem tractable, we propose a two-layer method: the Projected CVaR-tail selector (PCVaR-tail) for sparse lower-tail-aware user selection, followed by sequential least-squares quadratic programming for continuous placement and time allocation. Across ten traffic regimes, PCVaR-tail improves held-out global CVaR over the strongest Fixed-p heuristic by 3.1%, 10.1%, and 14.1% at α = 0.25, 0.10, and 0.05, respectively. Additional baselines, stress tests, high-load scenarios, and trajectory diagnostics show that PCVaR-tail provides a practical, interpretable lower-tail fairness mechanism for scheduled-bus WPCNs.

키워드