Streaming Task Research Articles

Placement Strategies for Internet-Scale Data Stream Systems

Optimally assigning streaming tasks to network machines is a key factor that influences a large data-stream-processing system's performance. Although researchers have prototyped and investigated various algorithms for task placement in data stream management systems, taxonomies and surveys of such algorithms are currently unavailable. To tackle this knowledge gap, the authors identify a set of core placement design characteristics and use them to compare eight placement algorithms. They also present a heuristic decision tree that can help designers judge how suitable a given placement solution might be to specific problems.

Dorsal and ventral stream mediated visual processing in genetic subtypes of Prader–Willi syndrome

Previous work has suggested that there are specific deficits in dorsal stream processing in a variety of developmental disorders. Prader-Willi syndrome (PWS) is associated with two main genetic subtypes, deletion and disomy. Relative strengths in visual processing are shown in PWS, although these strengths may be specific to the deletion subtype. We investigated visual processing in PWS using an adapted Simon task which contrasted location (dorsal stream) and shape identity (ventral stream) tasks. Compared to a group of typically developing children, children with PWS deletion showed a greater degree of impairment in the dorsal stream task than in the ventral stream task, a pattern similar to that shown in a group of boys with Fragile-X syndrome. When matched on a measure of non-verbal ability, children with PWS disomy showed the opposite pattern with better performance in the location compared to the shape task, although these task performance asymmetries may have been linked to executive control processes. It is proposed that children with PWS deletion show a relative strength in visual processing in the ventral stream along with a specific deficit in dorsal stream processing. In contrast, children with PWS disomy show neither effect.

Maximizing Total QoS-Provisioning of Image Streams with Limited Energy Budget

To fully utilize the limited battery energy of mobile electronic devices, we propose an adaptive adjustment method of processing quality for multiple image stream tasks running with widely varying execution times. This adjustment method completes the worst-case executions of the tasks with a given budget of energy, and maximizes the total reward value of processing quality obtained during their executions by exploiting the probability distribution of task execution times. The proposed method derives the maximum reward value for the tasks being executable with arbitrary processing quality, and near maximum value for the tasks being executable with a finite number of processing qualities. Our evaluation on a prototype system shows that the proposed method achieves larger reward values, by up to 57%, than the previous method.

Contributions of the brainstem and below to auditory scene analysis

We are increasingly aware of the cues the auditory system uses in the segregation and fusion of auditory objects, however, we have very little evidence for how, or where, these cues are extracted in the auditory pathway. Here I will show how the inner ear and cochlear nucleus contribute to three classical auditory scene analysis phenomena. In the first, grouping by common onset; adaptation, lateral suppression, and recovery from suppression all play a role. In the second phemonenon, a sequential streaming task, commonly referred to as ABA, the contributing low‐level mechanisms are frequency tuning and neural adaptation. Finally, in the third phenomenon, comodulation masking release, which describes the seemingly paradoxical situation whereby the detection of a masked signal is improved by the addition of coherently modulated energy remote from the signal frequency, lateral suppression is an important mediator. Taken together these studies suggest that neurophysiological processes at, or before, the level of the cochlear nucleus make an important contribution to auditory scene analysis.

Audio visual interaction in an auditory streaming task involving vowels

Previous studies suggest that sequential segregation mechanisms might be primary in concurrent speech segregation tasks for hearing‐impaired listeners. Other studies involving speech materials indicate that timbre cues and pitch cues primarily modulate sequential segregation. Finally, concurrent speech segregation has been reported to be enhanced by visual cues associated with speech reading. In this study, sequences of vowels having alternating fundamental frequencies (low and high) were presented to normal‐hearing listeners who were asked to report the presentation order of the vowels. Lower performance in such a task has been related to increased sequential segregation. By directing listeners to the perception of a single stream, this task leads to an estimation of the pitch‐based obligatory streaming evoked by the sequence. Simultaneous with the presentation of each high‐frequency vowel, visual cues simulating lip movements were provided. Lip movements were either steady (control condition), open‐closed (rhythm condition), or congruent with each particular vowel (congruent condition). As such, lip movements were used in this study as a potential distracter. The lower performance (increased streaming) in the congruent condition relative to control conditions suggest that visual cues may interact with sequential obligatory streaming mechanisms at a sensory level.

Stream Task Scheduling Method for Deadline-Sensitive Applications

Most of the existing real-time processing systems over data streams focus on minimizing average tuple latency while less attention has been paid to deadline of each individual tuple. This paper presents a real-time adaptive batch task scheduling (ATS) mechanism to support the strict deadline requirements of mission-critical applications over time-varying and bursting data streams. The ATS strategy aims at maximizing task throughput and minimizing deadline miss ratio by minimizing both scheduling overheads and deadline miss overheads. The paper proposes a concept of the optimal scheduling unit—batch granularity, and designs a closed-loop feedback control mechanism to adaptively select the dynamic optimal batch size in a non-predictable data stream environment. The theoretical analyses and experimental results show the efficiency and effectiveness of the ATS batching technique.

Decentralized media streaming infrastructure (DeMSI): An adaptive and high-performance peer-to-peer content delivery network

Hosting an on-demand media content streaming service has been a challenging task mainly because of the outrageously enormous network and server bandwidth required to deliver large amount of content data to users simultaneously. We propose an infrastructure that helps online media content providers offload their server and network resources for media streaming. Using application level resource diversity together with the peer-to-peer resource-sharing model is a feasible approach to decentralize the content storage, server and network bandwidth. Each subscriber is responsible for only a small fraction of such resources. Most importantly, the cost of maintaining the service can also be shared amongst subscribers, especially when the subscriber base is large. As a result, subscribers can be benefit from lower subscription cost. There have been a few solutions out there that focused only on sharing the load of network bandwidth by division of a streaming task to be carried out by multiple sources. However, existing solutions require that the content to be replicated in full and stored in each source, which is impractical for a subscriber as the owner of the storage resource that is of consumer capacity. Our solution focuses on the division of responsibility on both the network bandwidth and content storage such that each subscriber is responsible for only a small portion of the content. We propose a light-weighted candidate peer selection strategy based on avoidance of network congestion and an adaptive re-scheduling algorithm in order to enhance smoothness of the aggregated streaming rate perceived at the consumer side. Experiments show that the performance of our peer-selection strategy out performs the traditional strategy based on end-to-end streaming bandwidth.

Speech perception in gated noise: The effects of spectral and temporal resolution and auditory streaming

Hearing-impaired (HI) listeners often report difficulty understanding speech in the presence of background noise, even in the presence of mild degrees of hearing loss. In addition, HI listeners show significantly less benefit from fluctuations in noise than do normal hearing (NH) listeners. Furthermore, HI listeners often show less accuracy of processing frequency and temporal information in acoustic signals and grouping them into a whole meaningful speech. The purpose of this study is to examine differences in performances of NH and HI listeners for speech perception in various types of noise. Three hypotheses will be examined in this study. First, the amount of masking release in HI listeners with mild hearing loss will be smaller than in NH listeners even when speech and noise are presented at intensities sufficient to overcome the hearing loss. Masking release will be measured for consonant recognition and sentence recognition by subtracting the percent correct in steady noise from that in gated noise. Second, the performance of HI listeners on spectral resolution, temporal resolution, and auditory streaming tasks will be significantly poorer than that of NH listeners. Third, the performance differences in these tasks may account for differences in masking release.

Behavioral evidence for brain-based ability factors in visuospatial information processing

The present study examined possible parallels between the structure of human visuospatial abilities and the organization of the neural systems. Forty-eight participants were tested on seven speeded visuospatial tasks. Three of these tasks were constructed so as to rely primarily on known ventral stream functions and four were constructed so as to rely primarily on known dorsal stream functions. Both sets of tasks spanned approximately the same range of difficulty as indexed by both the speed and accuracy of decision making. Factor analysis of response times on the seven tasks revealed only two significant factors. The putative ventral stream tasks all loaded heavily on one factor (mean loading=0.843) but only weakly on the other factor (mean loading=0.222); the putative dorsal stream tasks showed the opposite pattern in that they all loaded heavily on the second factor (mean loading=0.828) but only weakly on the first factor (mean loading=0.229). These findings are consistent with the hypothesis that human visuospatial abilities can be classified using categories based on the specializations of underlying neural structures and systems.