UMP-PerComp: A Ubiquitous Multiprocessor Network-Based Pipeline Processing Framework for Pervasive Computing Environments

Kubo, Mitsunori; Ye, Baoliu; Shinozaki, Arata; Nakatomi, Takayuki; Guo, Minyi

doi:10.1109/aina.2007.142

Cited by 13 publications

(13 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our project constructed a highperformance simulation environment for heterogeneous multiprocessor system [8] to verify a basic framework for inter-processor communication. [9] presents an optimized inter-processor communication algorithm using a double-buffered model In the next step, UMP-PerComp [5] was proposed with basic communication protocol, which exploits both parallelism and pipeling to provide users with the full power of the available resources existing in ubiquitous processors. A JPEG encoding application is applied on a software model based on UMPPerComp, and the basic protocol is verified.…”

Section: Related Workmentioning

confidence: 99%

“…This output specification of the GP-PE differs from SS-PEs in some cases, which output results at the decided data granularity. The DispatchService thread calls (5) FinishTask to free the memory allocated for the task after it is finished. When a DPL is deleted from a GP-PE, the Server thread calls (6) UnInitializeDPL to free the memory such as the table InitializedDPL generated.…”

Section: Figure 3 Sequence Of Function Callmentioning

confidence: 99%

See 1 more Smart Citation

Development of General-Purpose Processing Element and Network-Based Dataflow Processing System: Part One

Sugahara¹,

Honda²,

Shinozaki³

et al. 2009

2009 Fourth International Conference on Frontier of Computer Science and Technology

Self Cite

View full text Add to dashboard Cite

This research concerns a system, which is called Ubiquitous Multi-Processor (UMP) system, which constructs dataflow network dynamically by connecting function-specific execution modules. The software model of the system was developed consisting only of application-specific execution modules (SS-PE) providing fixed functions. We developed a general-purpose execution module (GP-PE) which performs equivalent function to SS-PE by downloading the software library to improve the efficiency of development and reconfigure the function dynamically. The GP-PE was evaluated on the loading time of the library, the processing time of a color space conversion application, and the configuration time of processing on a network. The results of evaluation show that the GP-PE achieves equivalent performance to the SS-PE. Furthermore, the GP-PE does not degrade the system performance regarding dynamic system configuration if the library becomes hundreds of megabytes. This result shows that the GP-PE is suitable for the UMP system in general use.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Figure 3 Sequence Of Function Callmentioning

confidence: 99%

Development of General-Purpose Processing Element and Network-Based Dataflow Processing System: Part One

Sugahara¹,

Honda²,

Shinozaki³

et al. 2009

2009 Fourth International Conference on Frontier of Computer Science and Technology

Self Cite

View full text Add to dashboard Cite

show abstract

“…The proposed model showed over 50% system performance improvement rate compared with the basic framework integrated with the single-buffered model [5]. As the next stage of the hardware simulation, we implemented a ubiquitous multi-processor network based pipeline processing framework to support the development of high performance pervasive application [6]. In [6], we also developed a distributed JPEG encoding application successfully based on the proposed framework.…”

Section: Introductionmentioning

confidence: 96%

Performance Analysis of Resource Allocation Algorithms Using Cache Technology for Pervasive Computing System

Dong

Guo

Zheng

et al. 2008

2008 the 9th International Conference for Young Computer Scientists

Self Cite

View full text Add to dashboard Cite

With the great progress of the Information Technologies, micro-processors are embedded into everywhere to make our daily life convenient, efficient and comfortable. However, how to integrate distributed resources existing among heterogeneous devices to build pervasive applications that can constantly adapt to the highly dynamic computing environment is still a challenge. On our previous work, by analyzing the pipelining feature within a user task and exploiting the parallelism among the ubiquitous processors, we have proposed UMP-Percomp, which is a Ubiquitous Multiprocessor-based pipeline processing architecture to support high performance pervasive application development. As a case study of the UMP (Ubiquitous Multi-Processor) system, we have implemented a prototype system of the JPEG encoding. In this research, we have studied three different scheduling policies to the prototype system and proposed the best architecture of the UMP system. We have also introduced the cache technology to the proposed resource allocation algorithm. To prove our ideas, a simulation based system was implemented. The experience results revealed that the proposed algorithms made a significant progress over the existing solution.

show abstract

“…In order to evaluate the scalability and performance of the heterogeneous multiprocessor systems, a basic framework of multiprocessor simulation system has been implemented based on a multi-way cluster [5] and a doublebuffered communication model [6] has been incorporated into the system that can improve the performance over 50%, in terms of communication speed. We have extended the system and implemented a ubiquitous multi-processor network-based pipeline processing framework [7], at the hardware simulation level, to support the development of high performance pervasive applications. In addition, we succeeded in the development of a distributed JPEG encoding application based on the framework and devised the method in an efficient way [8][9] [10].…”

Section: Introductionmentioning

confidence: 99%

A probabilistic-approach based resource allocation algorithm in pervasive computing systems

Dong

Zheng

Ota

et al. 2010

2010 International Conference on Computer Application and System Modeling (ICCASM 2010)

Self Cite

View full text Add to dashboard Cite

Ubiquitous technologies are indispensable for modernizing human daily life more and more. However, the technologies are not easily widespread everywhere in our world through infrastructures and other related techniques. We have worked on a project to meet these challenges with a goal to construct a framework for the coming ubiquitous society. In our previous works, we have proposed UMP-PerComp, a Ubiquitous Multiprocessor-based pipeline Processing architecture, to support development of powerful and pervasive applications. In this paper, we propose an optimized algorithm for the UMP system to improve resource allocation executed by one kind of processing nodes called the Resource Router (RR). Using the optimized algorithm, the RR can effectively find a node in the idle state, which actually processes a task assigned by the RR. As a result, the RR saves the time to search for an idle node so that total performance can be improved. Finally, we evaluate the optimized algorithm with probability analyses to show effectiveness more than the previous algorithm we used.

show abstract

UMP-PerComp: A Ubiquitous Multiprocessor Network-Based Pipeline Processing Framework for Pervasive Computing Environments

Cited by 13 publications

References 10 publications

Development of General-Purpose Processing Element and Network-Based Dataflow Processing System: Part One

Development of General-Purpose Processing Element and Network-Based Dataflow Processing System: Part One

Performance Analysis of Resource Allocation Algorithms Using Cache Technology for Pervasive Computing System

A probabilistic-approach based resource allocation algorithm in pervasive computing systems

Contact Info

Product

Resources

About