Refactor And Optimization Of CORBA Data Server In AOI System

AOI(Automatically Optical Inspection) is a kind of quality verify system. It iswidely used on electronic device manufactory assembly lines, can diagnose thedefects on products during the manufactory process is on the fly. This thesis didresearch on an AOI system from author’s company. This system has5-6years historyand keeps in leadership position in global AOI field. But with the electronic devicesbecoming more and more complex, it’s required to add new features to originalsystem. Also, assembly lines are faster and faster, so the performance pressure oforiginal system is even higher. To support new features and improve performance, it’srequired to upgrade both software and hardware components in original system.This thesis focuses on a software component Defect Server (DFS) in originalsystem. DFS provides data service for other components like a database. It has veryhuge performance pressure in AOI system. The original DFS costs lots of time inanalyzing communication content, impacts system performance. And the originalcommunication protocol is difficult to support adding new components in new AOIsystem. So in new version of DFS, we use CORBA as communication protocol inDFS to improve expansibility and performance. This thesis including:1. Introduce an AOI system and DFS. Figure out the system weakness.Provide new design of the system, including new work process, newcomponents such as GPS and Judgment Server.2. To improve expansibility and performance of DFS, provide a new design of software structure and communication layer. Add a new CORBAcommunication layer besides original SOCKET layer.3. Design performance experiments for IDL, POA policy, asynchroniesinvocation. According to experiment results, discuss how to use thesetechnologies to improve DFS performance.4. Explain some real problems from our customer side such as conflictionbetween thread synchronies and performance. Discuss how to debug orfixing problems using various POA policies.5. Benchmark and test the DFS after refactor. Collect the data from customerside demonstrate the optimization of DFS meet the requirement.