The main thrust of my research activities at the University of Michigan - Dearborn (UM-D) is in the electrical and computer engineering field. The contributions are divided into two primary areas. They are (1) fuzzy logic engineering applications with emphasis on automotive applications; (2) Software Engineering; (3) Cloud Computing and (4) intelligent algorithms for memory management, arithmetic computation units and interconnected networks in parallel system architectures.
The main focus of my research in fuzzy logic is on the development of intelligent controllers for automotive applications. Fuzzy logic have gained the trust of international automotive industry as an intelligent design mathematical system after the success that the Japanese have had using it.
The past twenty years in automotive history have seen a profound performance enhancement. This is linked, on one hand, to the awareness of automobiles contributing to the catastrophic rise in environmental pollution, and on the other hand, to the continued search for drive comfort and safety in order to transform the automobile from a simple tool to an integral part of daily life. The implementation of severe regulations related to emission control, fuel consumption, safety, as well as the search for improved derivability has led to the need for precise control of vehicle operation and improved handling qualities. Therefore, today many vehicle functions are currently performed through strategies implemented in various types of electronic control units. However, conventional control methods used in these units are sometimes unsuccessful due to a lack of robustness, the necessity of complex process models, and elaborate tuning procedures due to the continuo variation of parameters.
Fuzzy logic based automatic braking system was proposed that uses distance and relative speed sensors as inputs and brake-pressure as an output. Heuristic rules were developed and implemented. The controller monitored the deceleration rate of the vehicle to prevent tire lock-up and the consequent loss of directional stability. The system offered the flexibility of setting the separation distance. Simulation of the controller for driving into stationary or moving objects showed that the system performed well. It also used an anti lock braking system to decelerate the vehicle and a throttle in-off controller to accelerate the vehicle and maintain a fixed separation distance behind the object in a tracking mode.
Automation of gasoline fuel-injection evaporative purge-calibration using fuzzy controller with a fuzzy self-tuning feature was also proposed, designed and simulated. The fuzzy controller automated the manual calibration process for the estimation of the purge vapors scale factor. A fuzzy self-tuning scheme, based on a gradient decent algorithm, was considered to enhance controller performance. Various computer simulations have successfully demonstrated the automatic and adaptive capabilities of the proposed fuzzy controller. In addition to good performance, this controller does not require any additional sensors when compared to existing controller.
A fuzzy model of the route choice process was developed based on the present production Automap Navigation System and then simulated using the CubiCalc fuzzy simulator package. The simulation ranked the route alternatives in the same order as the existing production navigation system. This model was then greatly simplified and re-simulated, with performance equal to the original model. The route choice model was then extended to include additional input variables not present in the Automap System and a possible scenario was simulated. The research concluded that for present automotive navigation systems, the use of fuzzy set theory merely complicated the system without improving performance. However, future navigation systems that use more input data may benefit from the use of fuzzy logic.
Other automotive applications using fuzzy logic have been researched such as fuzzy logic based controller for suspension system, automotive fuzzy airbags, vehicle steering fuzzy controller, and others.
Evaluation of performance of employees has been, for a long time, a major preoccupation of university researchers and management practitioners. Performance appraisal (PA) is a formal, structured system of measuring and evaluating an employee’s job-related attributes, behaviors, and outcomes to assess an employee’s productivity and judge whether he or she will perform as or more effectively in the future. Information from performance evaluation is used for four principal purposes: (1) between-person development (i.e., salary administration, promotion, and layoffs); (2) within-person development (i.e., performance feedback, identification of individual strengths/weaknesses, and identification of individual training needs); (3) systems maintenance (i. e., development of organization goals, human resource planning, and determination of organizational training goals); and (4) documentation (i.e., documentation of human resource decisions, meeting human resource legal requirements, and criteria for validation research).
The establishment of performance criteria is an important prerequisite to performance appraisal that provides employees with an understanding of the expected levels of accomplishment. However, what does supervisory identification of an "outstanding" performance mean? What does "poor" mean? What is "average"?. For jobs involving more than one task, there is another difficulty to be overcome concerning the combination of separate aspects of performance into a composite score that will facilitate interemployee comparisons. One way is to weight each criterion equally. It has been argued that effective performance evaluation systems evolve from recognition that human behaviors and capabilities that are collapsed into a single score have limited use in developing human resources.
The purpose of this research was to suggest an application of fuzzy logic modeling to personnel PA. In particular, the case evaluated in this research applies fuzzy logic to the analysis and evaluation of PA system at an academic department of a higher-education institution located in the Midwest region of the USA.
The state-of-the-art in fuzzy, neuro-fuzzy and intelligent appliance technology was presented in. Appliances of the future home will automatically adjust to several house-related factors such as the number of people present, temperature and light intensity levels, and even to the cleanliness of the floor. The appliances will even operate themselves. Fuzzy logic has helped bring these dreams to reality. It has permeated many aspects of life in general, especially in Japan and increasingly in the US. The concept of the fuzzy-controlled future home has already appeared in Japanese trade shows and households. Numerous domestic appliance applications use fuzzy logic to achieve design goals. One goal is that the appliance should be simple to operate. The second, that the appliance should have a short development time. The third, that the appliance should be cost effective compared to its standard logic counterparts. Finally, the designs should be dynamic, with the ability to adjust to new inputs and different users. Fuzzy logic has allowed designers to achieve all of these goals.
While there are many software processes currently available, they are usually focused on large projects with multiple programmers. True, there are processes geared towards one-man teams, however, there is still the assumption that the project in question is a larger one. This research introduces ResPCT, a software process geared toward the one-man team working on a small to medium sized project. ResPCT requires four phases, Research, Prototype, Create and Test. ResPCT was applied to various applications and showed a great success as a new software method.
Given the central role that software development plays in the delivery and application of information technology, managers have been focusing on process improvement in the software development area. This improvement has increased the demand for software measures, or metrics to manage the process. This metrics provide a quantitative basis for the development and validation of models during the software development process. In this research a fuzzy rule-based system will be developed to classify java applications using object oriented metrics. The system will contain the following features: Automated method to extract the OO metrics from the source code, Default/base set of rules that can be easily configured via XML file so companies, developers, team leaders, etc, can modify the set of rules according to their needs, Implementation of a framework so new metrics, fuzzy sets and fuzzy rules can be added or removed depending on the needs of the end user, General classification of the software application and fine-grained classification of the java classes based on OO metrics, and Two interfaces are provided for the system: GUI and command.
In this area of research many cloud applications are investigated. Repetitive visits to grocery stores and over-purchasing of grocery items is a common problem that people are faced with in USA. The idea of GroceryHub App is to allow individuals maintain and synchronize a shared list of items with people they share grocery with. GroceryHub is a free cloud-based solution that makes use of cutting technologies like Amazon S3 and Heroku. GroceryHub fully and flawlessly supports the offline mode in case a user doesn’t have network connectivity. It has the ability to attach images to items.
Another cloud application is the AlHajj. In today’s mobile world, more people have started to use smart-phones to consume information. Therefore usage of mobile apps and mobile websites for information consumption has increased. This research introduces AlHajj app for iOS which is an interactive guide to Hajj, like an interactive map allowing users to walk through the process of the hajj to develop a better understanding of the obligations, locations, dates and sequence they are performed in.
Another application is to use the Platform as a Service (PaaS) methodology and to utilize the cloud as a storage mechanism for personalization data that is commonly stored in consumer electronic devices such as radios, televisions, and other devices that commonly store presets or other configuration data. In generating the application for this research, several considerations for cloud based application development and deployments are identified, and should be considered for future cloud-based software developers and companies wishing to deploy such solutions.
The object of this research area is to develop parallel and efficient algorithms for load balancing in a distributed system, cache memory, and trigonometric functions.
Load balancing of individual processors in a distributed system has been a hot topic in recent years as the popularity of distributed systems has increased. In particular, algorithms to schedule and migrate tasks on processors have received a lot of attention. As processor speed increases and the number of processors in a system increases, load balancing becomes more important in order to utilize the system efficiently. One of the main problems encountered when trying to develop an effective algorithm to balance load over several processors is the fact that the "load" of a system is very ambiguous. What does the load of a system really mean? It is generally accepted that the run queue length of a processor is a good measure of how heavily loaded it is , but this is not necessarily the case – load can also depend on how much memory is available, rate of incoming tasks, etc. Because of the inherent ambiguity of processor load, fuzzy techniques are a good fit in the realm of load balancing. This research has presented a job-scheduling algorithm that utilizes fuzzy rule-based engine to balance load among processors to maximize the overall throughput and reliability of the system. The main focus of this research has been harnessing the simplicity of fuzzy to create a powerful load-balancing algorithm with a minimum amount of effort. In addition, the results presented in this research were taken from months of data collected in a production processing environment rather than artificial simulations. The fuzzy algorithm completed nearly twice the jobs as the non-fuzzy algorithm during the same sampling period.
Cache memories are used to improve processor performance by reducing the number of times the processor must reference main memory for data or instructions. A cache controller is used to make decisions on block replacements within a cache. The objective of this research has been to improve cache performance by using fuzzy logic in cache update policies. The measure of performance improvement is the hit ratio and execution time. The fuzzy cache hit ratio was compared to the FIFO, LRU, CIR and OPT replacement policies. The fuzzy cache replacement policy has proven to yield a more consistent hit ratio for various program types. The fuzzy cache learned with time and yielded the highest hit ratio compared to existing cache algorithms (excluding OPT). The fuzzy cache replacement policy was flexible, variables could be adjusted and linguistic values could be added and/or changed.
Memoryless algorithms for the evaluation of sinusoidal functions were developed. Differing from the MacLaurin power series method, the algorithms proposed in this research expressed an angle ? as a multiple A of ? thus ? = A?, where A is represented in binary, in a pipelined manner to calculate sinusoidal functions. Fully pipelined, this purely combinational approach has resulted in the evaluation of the sine and cosine, or arc sine and arc cosine, of an argument to be produced in parallel in a time interval approximately half that needed for traditional power series methods while requiring much less complicated circuitry. Furthermore, the hardware implementation of these algorithms resulted in an iterative cellular array, a property desirable to both VLSI design and expandability.
The objective of this research is to develop a fault-tolerant storage network capable of automatically managing and optimizing its reliability by monitoring and responding to failure indicators. This system is a fully automated self-healing network that is aware of its own behavior and failure profile, and is thus capable of proactively managing its own allocated storage units to minimize downtime and loss of information due to failure. We apply the latest research in disk failure model as a basis to implement new techniques for the construction of this fault-tolerant storage system.
Research using fuzzy logic in engineering applications with emphases in automotive applications and computer architecture will be my main future focus. Currently, I am working (with a graduate student) on job scheduling algorithms using adaptive fuzzy logic and genetic algorithm techniques to improve load balancing within a distributed system.
In the area of fuzzy automotive applications, I am working as a team leader on the following three automotive applications: airbag adaptive fuzzy controller, fuzzy logic based wheelchair system, and fuzzy control based consumer products. Also I am working on developing tools for fuzzy system testing.
I am also working in simulation and modeling hardware systems using VHDL.