DSpace Community: Faculty of Technology

A General Pyramidal Modular Neural Network Architecture for High Dimensional Input Vectors

Thu, 23 May 2013 07:45:29 GMT

Title: A General Pyramidal Modular Neural Network Architecture for High Dimensional Input Vectors

Authors: Menore, Tekeba

Abstract: In this thesis new modular neural network (MNN) architecture is proposed. The basic building blocks of the architecture are small multilayer feed forward networks trained using the Back propagation algorithm (BPA). The newly proposed MNN Architecture is called Pyramidal MNN (PMNN). It is called Pyramidal for the number of the modules that constitutes the layers of network relatively decreases from the input layer to the output layer. An Optimization technique called PSO has been used to optimize the topology of the proposed PMNN architecture for typical high dimensional input vector datasets. The optimization technique is used to suit the PMNN architecture for specific problems of high dimensional input vectors depending on the nature of the data input and the nature of the problem. This is done by evolving topology of the modules that constitutes the network and changing the architecture of the overall network to suit the new data set. The suggested training algorithm works in multiple stages depending on the number of hidden layers of the network. The training of modules in the same layer of the PMNN is easy to implement in parallel. Since the network is not fully connected, the number of weight of connections is less and hence the training is very quick for large input dimensional vectors. An object-oriented implementation of the proposed PMNN architecture is written to simulate the behavior. The evaluation and optimization of the PMNN architecture for different real world applications is carried out to show the effectiveness of the proposed architecture for high dimensional input vector applications. The evaluation is based on three pattern recognition problems: palm-print recognition, iris recognition and face recognition. In all the three evaluations, it has achieved more than 95% accuracy of the test results. Furthermore, the proposed PMNN architecture performs better than other similar type research works. It is shown that as PMNN is a huge family of several specific architectures, this proposed topology of the neural net can serve wide range of complex domain problems that need to be solved using Artificial Intelligence (AI).

Small Scale Metalworking Industries Cluster (Case Study on Towns of Oromia Special Zone)

Wed, 17 Apr 2013 08:54:02 GMT

Title: Small Scale Metalworking Industries Cluster (Case Study on Towns of Oromia Special Zone)

Authors: Dagne, Birhanu

Abstract: This paper deals with clustering of small scale metalworking industries in Oromia Special Zone (OSZ). Clustering is a group of small firms operating in a defined geographic location, producing similar products, cooperating and competing with one another, learning from each other in order to overcome internal problems, setting common strategies to overcome external challenges, and reaching distance market through developed networks. In order to attain benefits of cluster, industries are identified along with problems facing them by in depth interviews, discussions and questionnaires. These results are supported by literatures published and unpublished. The paper also addresses the background of OSZ with respect to industries. The industries are identified by the selective methodology with their performances based on number of workers, capital and value additions. Each location of industries is marked by Geographical Positions System and located the projections on Archi-Geographic Information System software in the world geographical positions. The industries are evaluated whether the naturally are following or not with their actual trends. The exact geographical location of a cluster for industries identified is placed by using gravity location systems on Microsoft Excel solver tool. By methods of model development for the cluster the model is developed and hence the growth paths of a cluster by evaluating value chain and stakeholders of industries that had direct or indirect linkages with industries for continuous growth of cluster. Besides, the internal processes are modeled using basic processes. With the specified principles of clustering, Ethiopian clusters are evaluated and argued in relation to infrastructure, technology, skilled manpower, information and e-commerce. The gap left by cluster- based strategies is filled by policy recommendations. The overall data and inputs are analyzed in relation to cluster performances are summarized in a meaningful manner. Trade-offs between clustering and scaling up in fast transferring of technology in Ethiopia is also discussed. The whole results and findings are tied under summary of the major findings. Finally, the paper ends with conclusions and suggesting the possible solutions to the problem for the gaps remained void in the paper.

Cooperative MIMO System for WiMAX Technology

Mon, 15 Apr 2013 07:54:26 GMT

Title: Cooperative MIMO System for WiMAX Technology

Authors: Samuel, Medhn

Abstract: Multipath fading is one of the primary factors that degrade the performance of wireless networks. One of the most powerful techniques to combat the effect of fading is by using multiple antennas that provides space diversity and spatial multiplexing; this technique is called Multiple-Input Multiple-Output (MIMO). Cooperative diversity, an alternative form of realizing MIMO, has been recently proposed to realize the diversity advantage in a distributed manner. Cooperative diversity exploits the broadcast nature of wireless transmission and creates a virtual antenna array through cooperating nodes. Although, prior research in cooperative diversity considers users equipped with single antenna, in practical scenarios users may be able to accommodate multiple antennas due to the recent advances in semiconductor industry. Hence, the primary purpose of this thesis is to model, and by using simulation, investigates the end-to-end performance of a cooperative diversity system employing multi-antenna at cooperating nodes; the intention is to simultaneously exploit the diversity gain offered by the cooperative diversity and multiple antennas. One contribution of this work is, we propose a cooperative MIMO system where the cooperation uses amplify-and-forward (AF) relaying strategy and the MIMO is based on Vertical Bell-Labs Layered Space Time architecture (VBLAST). The proposed system is to be used for WiMAX technology. Complexity of the cooperative MIMO system is also analyzed considering Maximum Ratio Combining (MRC) and Equal Gain Combining (EGC) schemes for Zero Forcing (ZF), Minimum Mean Square Error (MMSE) and Maximum Likelihood (ML) detection techniques. Simulation results show that cooperative MIMO system achieves significantly better symbol error rate (SER) performance than conventional cooperative diversity and MIMO systems with comparable complexity. As an example, for SER of 10-3, cooperative MIMO system using MMSE detection has 10 dB SNR gain over systems that employ cooperative diversity or MIMO only.

DSP BASED IMPELEMENTATION OF FIELD-WEAKENING ON SYNCHRONOUS MOTOR FOR HIGH SPEED OPERATION

Mon, 15 Apr 2013 07:47:38 GMT

Title: DSP BASED IMPELEMENTATION OF FIELD-WEAKENING ON SYNCHRONOUS MOTOR FOR HIGH SPEED OPERATION

Authors: HENOK, BEREHANU

Abstract: Permanent magnet motors have high efficiency and torque density compared to induction and conventional synchronous motors in general. However, in applications requiring wide speed range, the air-gap field, which is produced by the permanent magnets, can’t be controlled directly to control the back-emf of the motor for field weakening operation. In conventional synchronous machines the excitation current can be controlled to control the field and hence the speed. This thesis presents development of field weakening control for Permanent Magnet Synchronous Motor (PMSM) using demagnetizing component of stator current. The control is implemented using Texas Instruments’ TMS320LF2401A Digital Signal Processor (DSP). The control method presented relies on the field orientated control (FOC) together with a field-weakening operation. The field-weakening operation is required to extend the speed in the constant power range of operation, where there is no more voltage adjustment possible to increase the speed. The field oriented control of the motors has been modeled, simulated using MATLAB and Some of the modules of field oriented control are tested practically and results are given in this thesis. The results demonstrate that the speed of the PMSM can be extended in the constant power range for applications requiring wide speed range like in electric vehicle.