No abstract

In this study, we propose a novel server load balancing scheme. Server load balancing is known to be a critical mechanism in network-based information service. Most of existing schemes don't take servers' loadings into account, which might fail to make the loadings of all servers evenly and drive the server system to work on the brink of being over-loaded and/or out of function. The proposed scheme aims at preventing the occurrence of malfunction and saving the power consumption of the server system under a low loading while providing a better performance in response time. All connection requests are firstly distributed into one server until a pre-specified portion of its maximum allowed serving load is achieved. The following requests are then served by another server in the same way. The experimental results demonstrate the feasibility of the proposed scheme.

Pipeline is a fundamental parallel programming pattern. Mainstream pipeline programming frameworks count on data abstractions to perform pipeline scheduling. This design is convenient for data-centric pipeline applications but inefficient for algorithms that only exploit task parallelism in pipeline. As a result, we introduce a new task-parallel pipeline programming framework called Pipeflow. Pipeflow does not design yet another data abstraction but focuses on the pipeline scheduling itself, enabling more efficient implementation of task-parallel pipeline algorithms than existing frameworks. We have evaluated Pipeflow on both micro-benchmarks and real-world applications. As an example, Pipeflow outperforms oneTBB 24% and 10% faster in a VLSI placement and a timing analysis workloads that adopt pipeline parallelism to speed up runtimes, respectively.

In this paper, we present a low-cost driving assistant system which can provide the surrounding image of a vehicle in a bird's-eye view. By implementing a fast image stitching algorithm on an embedded platform, this system can generate the around-vehicle view in real time from four fisheye cameras mounted around the vehicle. Moreover, we also propose a dynamic boundary determination method to allow the automatic switch among image sources and to reduce the chance of missing obstacles in the overlapping area. With this bird's-eye view surrounding monitoring system, drivers can visually perceive the condition of the surrounding environment. Driving and parking can then become more easily and safely.

In this paper, image processing techn to the analysis of near-infrared videos. The human activities in the videos. For detecting hum implement the Gaussian mixture modeling (GM background model and to perform foregr Additionally, we pay attention to commonly use equipments used in nighttime environment illumination and shadowing phenomena. Acc mode GMM is proposed which separately const GMM for different lighting conditions and switc by event detection. In order to cope with exc phenomenon, an efficient way of searching fo scan-lines is proposed to remove human shadow approach will provide reasonable bounding box human regions as detection results which will b a nighttime surveillance system.