Among the noteworthy technical capabilities highlighted in the "Technology Innovation Report" for the State of Hawaii is the Laboratory of Intelligent and Parallel Systems (LIPS) at the College of Engineering of the University of Hawaii. This laboratory, developed by Professor D. Y. Y. Yun over the past ten years, provides a broad scope of advanced research and development expertise in the engineering of software, systems and applications using parallel computing and distributed networking. The unifying theme of a "virtual laboratory" is emerging in the form of technology integration and delivery of application services based on remote access of mobile/distributed data and computing resources through the Internet (or other high speed networking) with 2D graphics or 3D visualization for the end user anywhere and any time. The work at LIPS integrates data acquisition devices with on-demand communication links to intelligent decision software and sophisticated capabilities to enable timely delivery of needed services to remote users. Four funded projects, exemplifying this concept of networked access of remote computations and intelligent decisions, are summarized as follows:
The following activities are representative of additional LIPS experience and capabilities:
Image Intelligence: Pioneering research and technology development at LIPS encompass a broad spectrum of on-going activities in 3D image processing and object modeling. The algorithm and software developed at LIPS provide an innovative solution to the surface mesh generation problem that achieves guaranteed quality of fit while increasing computational speed and minimizing storage/transmission requirements. The 3D triangulation system using a constrained resource management approach to obtain a near optimal polygonal surface approximation of 3D objects, currently undergoing the patent application process, provides the underlying core technology. Several categories of noteworthy derived or related technologies reveal the breadth and vitality of this R&D agenda. 3D Graphical Visualization: LIPS is equipped with the state-of-the-art 3D graphical visualization and rendering hardware and software. Image Compression: The Lab has active ongoing research in various image compression techniques. A NASA funded research project is involved in parallel image compression techniques based on neural networks and adaptive vector quantization. Content Search of Image Database: Research and development in real time search of massive databases has resulted in a new associative computing model based on optical holographic principles, which enables inherently parallel search into thousands of image databases in logarithmic time. This technique also has the capability of applying cognitive focus during associative search. Content-based retrieval from image databases, target recognition in radar images, tumor detection from X-ray, and remote sharing of satellite imagery are just a few applications of this technology.
Constrained Resource Planning: During the past ten years, LIPS has developed a powerful technique for planning, scheduling and optimization, known as Constrained Resource Planning (CRP). CRP is both a general methodology for guiding any heuristic problem-solving process and a stand-alone executable engine for resource management under tight constraints. CRP has been applied to solve more than 40 NP-complete/hard problems with remarkable solution quality and efficiency. Well-known difficult problems successfully solved by CRP include Traveling Salesman, Job-shop Scheduling, 3D Packing, 3D Model Recognition, Multiprocessor Scheduling, and Subgraph Isomorphism. These problems illustrate CRP’s capacity in handling any of the six broad application patterns: Production Optimization; Space Utilization; Occupancy Planning; Inventory Distribution; Facility Reservation; and Work-Shift Scheduling. Current research impetus lies in two directions: (1) to expand this powerful AI technique for even more difficult problems (such as Maximum Common Subgraph, Distributed Network Management, and 3D Volume Visualization) and (2) to develop a theoretically sound argument for the solution optimality and computational efficiency of CRP in general (rather than problem by problem).
Parallel and Distributed Computing: The Lab is also involved extensively both in high performance parallel and distributed computing research and in the application of supercomputing resources for simulation and visualization. LIPS has developed a dynamic multi-processor scheduler which simultaneously optimizes processor utilization and job completion time. Also, in the area of fault- tolerant computing, LIPS has developed a new adaptive fault-tolerant algorithmic approach for improving the computational stability of matrix and parallel linear system computation, which currently is the best parallel pivoting scheme.
Researchers & Technical Staff:
This website is currently maintained by Shailesh of LIPS. Send your comment and inquiry.