Please use this identifier to cite or link to this item: http://hdl.handle.net/10397/86790
Title: A concurrency integrity model for distributed product data management
Authors: Chan, Edmond Cheuk-kit
Degree: Ph.D.
Issue Date: 2010
Abstract: In today's manufacturing environment, enterprises having work groups geographically dispersed are not uncommon. In addition, different tasks of the product lifecycle are distributed at different geographic locations. A product data management (PDM) system is therefore required for controlling the distribution and maintaining the integrity of the product data throughout its entire life cycle. Multiple accesses to the system residing on multiple sites will cause concurrency problems. It is crucial that concurrency control must be provided to protect the data against a variety of possible threats, in particular, data inconsistencies must be avoided and relationships between data must be maintained. In particular, the current PDM technology is not completely suitable for PDM in distributed manufacturing environment. Thus, this research aims to develop a foundation in concurrent engineering support for distributed product data management (DPDM). This research begins with reviewing the suitability of existing database management systems for product data within the scope of data architecture. These systems are based on the technology for managing business data. In other words, they are not efficient when employed to handle product data. Therefore, an ontological approach for representing product data is proposed to describe the relationships between all the objects within a DPDM system. To secure the consistency of a DPDM system, the functions of the system must be error-free. A DPDM system specification is necessary. Firstly, a graphical representation model is developed to express the PDM functions. UML (Unified Modeling Language) sequence diagrams are used to model the actions of each of the functions and their interactions between users and the system. Temporal logic is then used to construct the formula of these functions. The model is further extended to represent DPDM functions.
Given that the traditional concurrency control methods were not purposely developed to meet the need of DPDM, the requirements for data storage and manipulation for DPDM systems are not well supported. Therefore, a concurrent control method that caters for version management and product architecture in DPDM is proposed. This research demonstrates how granularity and versioning are incorporated into a lock-based concurrency control model. The concurrent accessibility of an example product data is explained to illustrate the adjustability according to the actions taken by the users and the architecture of the corresponding data object. Locking is one of the well-known concurrency control techniques and more likely to be encountered in practice. Although lock-based concurrency control methods guarantee serializability of data access, the systems have the risk of deadlock as the transactions may wait for unavailable locks. An integer programming based mathematical model employing transaction scheduling is proposed to prevent the threat of deadlock by controlling transaction executions in DPDM system while efficiency is maintained. To validate the performance of the proposed methods, the strict two-phase locking (2PL) method, the two-phase granularity version (GV) locking method, the 2PL transaction scheduling method, and the hybrid GV transaction scheduling method are evaluated through simulation experiments. Read and Write accesses to composite objects are used to illustrate the comparisons between the models. Their performances are evaluated by comparing the number of late transactions. It shows that the GV locking model and the transaction scheduling model are better than the strict PL method while the hybrid model can substantially improve the concurrency of DPDM system.
Subjects: Hong Kong Polytechnic University -- Dissertations
Product management -- Data processing
Concurrent engineering
UML (Computer science)
Ontologies (Information retrieval)
Pages: xv, [160, 14] leaves : ill. (some col.) ; 31 cm.
Appears in Collections:Thesis

Show full item record

Page views

43
Last Week
1
Last month
Citations as of Mar 24, 2024

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.