LegalAct Distribution is an electronic collection of laws and regulations documents, official opinions, case law studies, the judicial practices as well as the model of legal acts of European legislation and other foreign rights, tariffs and similar documents of interest to companies in order to legally conduct their businesses. The solution is the system consisted of legal database as central storage place for all legal documents and several different applications with main purpose to provide fast search, easy access, visibility and up-to-date documents for a customer.

There are 4 realizations of this subsystem:

1. Web application hosted on application server. This implementation allows customers to access data through the web site from the browser. See http://www.lexonline.paragraf.rs  
2. Mobile application is implementation customized for the access from smartphones. It communicates with application server which serves as a business layer, providing requested data. Android app can be downloaded from Google play store  
3. Single (standalone) version is Java Desktop application for individual users, which has own database and enables offline mode of work and synchronization with latest updates over Internet or by CD. 
4. Network version is a client-server implementation of the system widely used within company LAN, but also enabled to work over Internet. This solution has two models of implementations and deployment:

• thin client implementation, where the server side of the subsystem contains a full local database and repository, which may be updated over Internet or by CD, whereas the client side contains only the application communicating with local database directly. This is the best solution for large companies due to its speed and less requirements for processing on the client machine.

• fat client implementation, where the server side does only dispatching of the update packages to the client applications which has own database. This solution makes the server side very simple and minimizes necessary network traffic as well.

  Please visit website http://paragraf.rs

mBid24 is client-server application that allows using of eBay database on mobile device. This project consists of two main applications rich client application and server application. Client application is written in Micro Java (J2ME) so it is easily ported on many different mobile device platforms (Nokia, SonyEricsson, Motorola, and Siemens etc). Server application is also written in Java SE, using servlet, JSP and JSTL technology.

I was Lead responsible for server application. My job included detailed design and implementation of complete server solution. There were four main interfaces are made:

1. Web interface that allows user registration and user account management, also this interface included site implementation for user support (please see http://www.bid-24.de)
2. Web interface that allows administrative account management, this interface represents content management system CMS. This is not public interface but only for administrative purposes.
3. Interface toward mobile phone can be considered SOAP like interface toward rich client application. This interface presents essence of the mBid24 service becaouse handle all necessery communication with mBid24 rich client application on the mobile phone. This interface is tested for support of vast amount of traffic, also it is optimized to ensure a minimal response time, so this featuers makes it very stable and reliable interface.
4. Proxy interface handles cryptography with client application to ensure user security. This proxy is converting internally crypto messages and connecting using HTTPS on eBay application. Proxy application is also used as filter to optimize GPRS traffic toward the mobile phone, so indiract costs of using this application is minimazed (GPRS traffic)

  Please visit website and try our exciting application at http://www.zesium.com

Apr 04 - Nov 04

Team member responsible for HTTP, RTP and RTCP components. The PoC Client 1.0 has been developed in conformance with the IETF, 3GPP, OMA (NEMS) and Siemens Specification documents. The PoC defines three communication channels (SIP, HTTP and RTP) and several applications: PoC Call Control (SIP, RTP/RTCP), Presence (SIP), Contact Management (HTTP), Chat (SIP) and Instant Messaging (SIP). The project is written in Embedded C++ using conventions for achieving the portability. The project has been developed in cooperation with the Zesium mobile, Germany.

1. Down sampling with integer factor - sampling rate is decresed with integer factor. Application shows the hardware solution and responses in time and frequecy domains.
2. Upsamping with integer factor - sampling rate is incresed with integer factor. Application shows the hardware solution and responses in time and frequecy domains.
3. Conversion with rational factor - sampling rate is changed by rational factor. Application shows the hardware solution and responses in time and frequecy domains.
4. Ad Hoc filter design technique - this technique for digital filter design is user frendly. Just by placing the zeros and poles into z-plane you are able to see the frequency response of that filter.

Purpose of this project was to make a cross reference of the noise removal and edge detecting filtars in a image processing. Also the task was to make a parallel and to compare every filtar to FIRE filtar that uses fuzzy logic. This project conssisted of several stages.
First stage was to make af design and implementation of the FIRE Fuzzy filter for filtering images corrupted by noise as reference to already proposed solution. Design, Implementation and tests are done in MATLAB.
Second stage is to improve fuzzy engine base by introducing some learning mechanism (algorithm) for it's fuzzy base gemeration, the Genetic algorithm was design and implemented. That implementation is done according to the theory of the learning algorithms.
Next stage was to investigate possibilty to use these fuzzy filters in edge detecting. Task was to design and implement a filters for edge detection based on fuzzy logic. Edge detection is performed on noise-corrupted images, so this filters uses FIRE filters as noise removers.