Opportunistic Slack Reservation Queueing (OSRQ): A Momentum-Driven Adaptive Scheduling Policy for Mixed-Priority Service Systems with Applications in Power, Renewable Energy and Irrigation

This paper proposes Opportunistic Slack Reservation Queueing (OSRQ), a continuous-time, single-server scheduling policy that enhances classical non-preemptive priority queueing through a demand-driven reservation mechanism. The policy leverages a sliding-window arrival estimator M(t) to capture short-term workload momentum. When M(t) exceeds a predefined threshold T, the scheduler temporarily reserves a service slot for a lead time Δ; if no urgent job arrives within this interval, the reserved capacity is efficiently reclaimed via backfilling by normal jobs, ensuring work conservation.

OSRQ is particularly suited for non-preemptive service environments with autocorrelated arrivals, including database workload management, serverless computing platforms and Kubernetes-based scheduling, where preemption is either infeasible or operationally costly. We formulate the problem as a continuous-time Markov Decision Process (MDP) on an augmented state space (Q_U, Q_N, M), establish system stability via Foster-Lyapunov drift conditions (ρ < 1) and demonstrate a threshold-optimal policy structure using supermodularity arguments.

A validated discrete-event simulation framework, benchmarked against M/M/1 and Cobham’s priority results, shows that under Markov-modulated Poisson process (MMPP) arrivals, OSRQ achieves a 1-3% reduction in mean urgent waiting time (WU) relative to the non-preemptive priority baseline, at a 12-20% increase in mean normal waiting time (W_N). Under Poisson arrivals, this benefit disappears, confirming that performance gains arise specifically from arrival autocorrelation, thereby validating the leading-indicator hypothesis. While preemptive priority scheduling yields superior WU, it does so at a significantly higher W_N cost. Overall, OSRQ provides a tunable intermediate operating regime between strict priority and static reservation, offering practical advantages in systems where preemption is constrained.

Beyond cloud and database workloads, OSRQ is shown to satisfy the non-preemptive, autocorrelated, cost-asymmetric fit criteria in three cyber-physical infrastructure domains: smart-grid SCADA and demand-response dispatch, battery energy storage and renewable inverter control and automated irrigation and canal gate management. Domain-specific momentum proxies: grid frequency rate-of-change, inverter output variance and hydrological event counts, map directly onto the OSRQ sliding-window estimator, broadening the scope of OSRQ from cloud-native scheduling to safety-critical physical infrastructure where the cost asymmetry between urgent and normal traffic classes is amplified by real-world consequences including cascading outages, regulation non-compliance and crop damage.