Impact Of Prediction Error Research Articles

Improving the Performance of Online Bitrate Adaptation with Multi-Step Prediction Over Cellular Networks

Video streaming over mobile is flourishing, and most commercial players use adaptive bitrate (ABR) streaming to deliver video in varying network conditions. Using network capacity and buffer occupancy as system states, ABR algorithms adjust bitrate based on the instantaneous system states, which is able to adapt to network changes in real-time and ensure high quality of experience (QoE). However, they are incapable of providing good QoE over mobile. Due to the high dynamic characteristics of cellular network, the system states change rapidly over time. The instantaneous state-based adaptation can induce significant video quality fluctuation which greatly degrades QoE. In this paper, we propose an online ABR algorithm called MSPC to provide good QoE in cellular network. To balance the conflict between rapid adaptation and smooth bitrate, MSPC utilizes the multi-step prediction of future system states to select bitrates instead of the instantaneous current states. At the same time, it controls the buffer occupancy to eliminate the impact of prediction error on performance. We implement MSPC on a reference video player with performance evaluated based on realistic cellular traces. Experimental results show that MSPC reduces the bitrate change of existing online algorithms by 62.4 percent on average while maintaining high bitrates and achieving zero rebuffering over 97.83 percent of all tested sessions.

Beyond arousal: Prediction error related to aversive events promotes episodic memory formation.

Stimuli encoded shortly before an aversive event are typically well remembered. Traditionally, this emotional memory enhancement has been attributed to beneficial effects of physiological arousal on memory formation. Here, we proposed an additional mechanism and tested whether memory formation is driven by the unpredictable nature of aversive events (i.e., aversive prediction errors). In a combined Pavlovian fear conditioning and incidental memory paradigm, participants saw initially neutral pictures from 2 distinct categories, 1 of which was associated with a risk to receive an electric shock. During encoding, we measured both physiological arousal and explicit prediction errors to explain memory differences in a surprise recognition test that followed approximately 24 hr later. In a first experiment, we show that physiological arousal, expressed as outcome-related skin conductance responses, was associated with improved recognition memory, corroborating arousal-based models. Critically, unsigned binary prediction errors derived from explicit shock expectancy ratings in each trial were also linked to enhanced recognition and model fits showed that the impact of prediction errors on memory was dissociable from the impact of arousal. In a second experiment, we replicated and extended the findings of the first experiment by demonstrating that the memory-promoting effect of prediction errors remained even after controlling for arousal. The present data point to prediction error-related learning as a cognitive mechanism that contributes to the emotional enhancement of memory, above and beyond the well-established effects of arousal in emotional memory formation. (PsycINFO Database Record (c) 2020 APA, all rights reserved).

The Impact of Prediction Errors in the Domestic Peak Power Demand Management

In this article, the impact of prediction errors on the performance of a domestic power demand management is thoroughly investigated. Initially, real-time peak power demand management system using battery energy storage systems (BESSs), electric vehicles (EVs), and photovoltaics (PV) systems is designed and modeled. The model uses real-time load demand of consumers and their roof-top PV power generation capability, and the charging-discharging constraints of BESSs and EVs to provide a coordinated response for peak power demand management. Afterward, this real-time power demand management system is modeled using autoregressive moving average and artificial neural networks-based prediction techniques. The predicted values are used to provide a day-ahead peak power demand management decision. However, any significant error in the prediction process results in an incorrect energy sharing by the energy management system. In this research, two different customers connected to a real-power distribution network with realistic load pattern and uncertainty are used to investigate the impact of this prediction error on the efficacy of an energy management system. The study shows that in some cases the prediction error can be more than 300%. The average capacity of energy support due to this prediction error can go up to 0.9 kWh, which increases battery charging-discharging cycles, hence reducing battery life and increasing energy cost. It also investigates a possible relationship between environmental conditions (solar insolation, temperature, and humidity) and consumers' power demand. Considering the weather conditions, a day-ahead uncertainty detection technique is proposed for providing an improved power demand management.

Dynamic Service Placement for Mobile Micro-Clouds with Predicted Future Costs

Mobile micro-clouds are promising for enabling performance-critical cloud applications. However, one challenge therein is the dynamics at the network edge. In this paper, we study how to place service instances to cope with these dynamics, where multiple users and service instances coexist in the system. Our goal is to find the optimal placement (configuration) of instances to minimize the average cost over time, leveraging the ability of predicting future cost parameters with known accuracy. We first propose an offline algorithm that solves for the optimal configuration in a specific look-ahead time-window. Then, we propose an online approximation algorithm with polynomial time-complexity to find the placement in real-time whenever an instance arrives. We analytically show that the online algorithm is $O(1)$-competitive for a broad family of cost functions. Afterwards, the impact of prediction errors is considered and a method for finding the optimal look-ahead window size is proposed, which minimizes an upper bound of the average actual cost. The effectiveness of the proposed approach is evaluated by simulations with both synthetic and real-world (San Francisco taxi) user-mobility traces. The theoretical methodology used in this paper can potentially be applied to a larger class of dynamic resource allocation problems.

Pricing data center demand response

Demand response is crucial for the incorporation of renewable energy into the grid. In this paper, we focus on a particularly promising industry for demand response: data centers. We use simulations to show that, not only are data centers large loads, but they can provide as much (or possibly more) flexibility as large-scale storage if given the proper incentives. However, due to the market power most data centers maintain, it is difficult to design programs that are efficient for data center demand response. To that end, we propose that prediction-based pricing is an appealing market design, and show that it outperforms more traditional supply function bidding mechanisms in situations where market power is an issue. However, prediction-based pricing may be inefficient when predictions are inaccurate, and so we provide analytic, worst-case bounds on the impact of prediction error on the efficiency of prediction-based pricing. These bounds hold even when network constraints are considered, and highlight that prediction-based pricing is surprisingly robust to prediction error.

Linking the Market and Internal Values of Jobs

The point factor method of job evaluation is one of the most popular and enduring approaches to linking the market and internal value of jobs. Statistically, regression analysis is used to create a market line that allows the organization to predict the market value of its jobs using point scores that define the jobs’ internal values. While attention is given to the fact that the market line represents the statistically “best” option for predicting these market rates, overlooked is the fact that these predictions almost always differ from the jobs’ actual market rates. This article explores the impact of this prediction error in compensation planning. It uses data on 41 jobs to define a market line using simple regression and identify the errors associated with the line’s predicted market values. It provides methods for precisely defining the extent of this prediction error and for minimizing it. It also discusses the impact of this error on the interpretation of salary grades, and the need for policy on key compensation planning issues to minimize the negative impact of prediction error.

An energy optimization strategy for power-split drivetrain plug-in hybrid electric vehicles

To demonstrate the greater capabilities and benefits achievable with a plug-in hybrid electric vehicle (PHEV), an energy optimization strategy for a power-split drivetrain PHEV, which utilizes a predicted speed profile, is presented. In addition, the paper reports an analysis and evaluation of issues related to real time control implementation for the modeled PHEV system, which include the optimization window sizes and the impact of prediction errors on the energy optimization strategy performance. The optimization time window sizes were identified and validated for different driving cycles under different operating modes and total length of travel. With the identified optimization windows size, improvements in fuel consumption were realized; the highest improvement was for Urban Dynamometer Driving Schedule (UDDS), with a range of improvement of 14–31%, followed by a 1–15% range of improvement for Highway Fuel Economy Driving Schedule (known as HWFET) and a 1–8% range of improvement for US06 (also known as Supplemental Federal Test Procedure). While no correlation was observed between the error rate and the rate of increased fuel consumption, this PHEV system still yielded energy savings with errors in the speed prediction, which is an indication of robustness of this PHEV model.

Impact of Uncertainty on the Prediction of Airspace Complexity of Congested Sectors

The ability of traffic controllers to separate aircraft determines the capacity of the region of airspace under their control, referred to as a sector. Complexity metrics, specifically dynamic density, is used as an estimate for controller workload. The prediction of dynamic density is required for the development of efficient strategic air traffic flow plans. This paper explores the influence of trajectory errors and unexpected aircraft on the prediction of dynamic density. A worst-case analysis is used to describe the conditions under which forecast uncertainty may lead to excessive complexity. Although the approach has general applicability, it is described using one complexity metric. Depending on the sector and the complexity function, when a sector is highly congested, the method identifies aircraft entering the sector at certain locations, boundaries, and altitudes, whose errors in prediction impact the increase in workload significantly. Results based on the analysis of 72 days of traffic data in the Atlanta Center show that the impact of prediction errors on complexity depends on the direction and altitude of the traffic. The normalized increase in complexity due to prediction errors in Sector ZTL50 is two times or higher for traffic coming from ZTL42 as compared with traffic coming from ZTL32. The analysis is repeated for Atlanta Sector ZTL06 and Washington Center Sector ZDC37 to show the applicability of the method to other sectors. For certain types of incoming traffic, if these errors cannot be reduced, it may be necessary to limit the traffic approaching the sector from these boundaries and altitudes.

A case for exploiting self-similarity of network traffic in TCP congestion control

Analytical and empirical studies have shown that self-similar traffic can have detrimental impact on network performance including amplified queuing delay and packet loss ratio. On the flip side, the ubiquity of scale-invariant burstiness observed across diverse networking contexts can be exploited to better design resource control algorithms. In this paper, we explore the issue of exploiting the self-similar characteristics of network traffic in TCP congestion control. We show that the correlation structure present in long-range dependent traffic can be detected on-line and used to predict the future traffic. We then devise a novel scheme, called TCP with traffic prediction (TCP-TP), that exploits the prediction result to infer, in the context of AIMD steady-state dynamics, the optimal operational point at which a TCP connection should operate. Through analytical reasoning, we show that the impact of prediction errors on fairness is minimal. We also conduct ns-2 simulation and FreeBSD 4.1-based implementation studies to validate the design and to demonstrate the performance improvement in terms of packet loss ratio and throughput attained by connections.