Timeout Policies Research Articles

A-11 Fatigue and Errors in Aeromedicine (FEAr) Pilot Study

Abstract Objective The FEAr pilot study was conducted to evaluate the construct validity of an instrument that measures self-reported fatigue levels among air medics (the Flight Risk Assessment [FRA] measure). Data Selection The population of interest, flight medics, is highly specialized and relatively small. As a result, non-probability, convenience sampling was used (N = 27). The participants were recruited from one air medical company with bases located across ten western states. Medics took two measures, the FRA and the PVT, three times a day, for three days. Data Synthesis To establish construct validity, scores on the FRA were compared to scores on the Psychomotor Vigilance Task (PVT), which is a widely used and validated measure to detect fatigue in field settings. The results of the study indicate that a total of five questions on the FRA was significantly and moderately correlated with the PVT efficiency score (r = -.49, p = .01). The mean reaction time was also significantly correlated with the total FRA score (r = .59, p = .001). Notably, perceived fitness for duty was poorly correlated with fatigue levels as measured by the FRA (r < .01). Conclusions The results of this study contribute to a very limited body of research on fatigue among air medical providers. Specifically, the study contributes to the knowledge about the construct validity of the Flight Risk Assessment, which can be used to evaluate fatigue among air medics. Further analysis can assist in determining appropriate cutoff scores to inform time-out policies for medics.

Online learning of timeout policies for dynamic power management

Dynamic power management (DPM) refers to strategies which selectively change the operational states of a device during runtime to reduce the power consumption based on the past usage pattern, the current workload, and the given performance constraint. The power management problem becomes more challenging when the workload exhibits nonstationary behavior which may degrade the performance of any single or static DPM policy. This article presents a reinforcement learning (RL)-based DPM technique for optimal selection of timeout values in the different device states. Each timeout period determines how long the device will remain in a particular state before the transition decision is taken. The timeout selection is based on workload estimates derived from a Multilayer Artificial Neural Network (ML-ANN) and an objective function given by weighted performance and power parameters. Our DPM approach is further able to adapt the power-performance weights online to meet user-specified power and performance constraints, respectively. We have completely implemented our DPM algorithm on our embedded traffic surveillance platform and performed long-term experiments using real traffic data to demonstrate the effectiveness of the DPM. Our results show that the proposed learning algorithm not only adequately explores the power-performance trade-off with nonstationary workload but can also successfully perform online adjustment of the trade-off parameter in order to meet the user-specified constraint.

Adaptive workload driven dynamic power management for high performance computing clusters

With the scale expansion of high performance computer systems, efficient power management has developed into an important issue. To strive to balance power consumption and performance, this paper proposes an adaptive workload-driven dynamic power management policy for homogeneous clusters, which dynamically adjusts the power mode of computing nodes according to workload variation. The proposed policy combines the pre-wakeup method and the feedback mechanism to reduce performance degradation due to the wakeup delay. The experimental results demonstrate that, as compared with two existing timeout policies, adaptive workload-driven dynamic power management effectively reduced the performance loss with a slight increase in power consumption.

MatchTree: Flexible, scalable, and fault-tolerant wide-area resource discovery with distributed matchmaking and aggregation

This paper proposes a novel wide-area resource discovery method, MatchTree, that is built upon a Peer-to-Peer (P2P) framework to deliver scalable and fault-tolerant resource discovery supporting distributed query processing and aggregation of results. MatchTree leverages a self-organizing tree for query distribution and result aggregation with the asymptotic latency increase pattern of O(logN), where N is the number of queried nodes. MatchTree distinguishes itself from related resource discovery systems based on structured P2P overlays by supporting complex queries (such as regular expressions in matchmaking), and from related unstructured P2P discovery systems by guaranteeing query completeness. This paper presents the overall architecture of MatchTree, proposes heuristics to improve fault-tolerance and reduce query response times through redundant query topologies, dynamic timeout policies, and sub-region queries. The paper evaluates the system quantitatively through large scale simulations, as well as through experiments with a prototype implementation deployed on a wide-area infrastructure (PlanetLab). The experiment results with proposed heuristics show that the maximum query latency of MatchTree decreases from 154 to 12 s, and the maximum query missing region decreases from 13.4% to 2.3% in the wide-area real world testbed.

Achieving a Vanishing SNR Gap to Exact Lattice Decoding at a Subexponential Complexity

This study identifies the first lattice decoding solution that achieves, in the general outage-limited multiple-input multiple-output (MIMO) setting and in the high-rate and high-signal-to-noise ratio limit, both a vanishing gap to the error performance of the exact solution of regularized lattice decoding, as well as a computational complexity that is subexponential in the number of codeword bits and in the rate. The proposed solution employs Lenstra-Lenstra-Lovász-based lattice reduction (LR)-aided regularized (lattice) sphere decoding and proper timeout policies. These performance and complexity guarantees hold for most MIMO scenarios, most fading statistics, all channel dimensions, and all full-rate lattice codes. In sharp contrast to the aforementioned very manageable complexity, the complexity of other standard preprocessed lattice decoding solutions is revealed here to be extremely high. Specifically, this study has quantified the complexity of regularized lattice (sphere) decoding and has proved that the computational resources required by this decoder to achieve a good rate-reliability performance are exponential in the lattice dimensionality and in the number of codeword bits, and it in fact matches, in common scenarios, the complexity of ML-based sphere decoders. Through this sharp contrast, this study was able to, for the first time, rigorously demonstrate and quantify the pivotal role of LR as a special complexity reducing ingredient.

Adaptive optimisation of timeout policy for dynamic power management based on semi-Markov control processes

Timeout policy is an industry standard for dynamic power management (DPM), and thus is easy and safe to implement in many power-managed systems. The optimisation of timeout policy suffered from the lack of effective analytical model and fell in heuristic previously. This study presents an adaptive optimisation method for timeout DPM policy. First, a semi-Markov control processes model is introduced to formulate the DPM problem of finding timeout policies that minimise power consumption under performance constraints. Under this framework, the equivalence of timeout and stochastic policies on power-performance tradeoff is revealed, and the equivalent relation between these two types of DPM policy is derived. Then, a reinforcement learning algorithm that combines policy gradient estimate and stochastic approximation is proposed for optimising timeout policy online. This algorithm does not depend on any prior knowledge of system parameters, and can achieve a global optimum with less computational cost. Simulation results demonstrate the analytical results and the effectiveness of the proposed algorithm.