GanapathyRaman

MafiaMania

Spam-Detector

The project is about developing and implementing an algorithm which can classify the mails as a spam or a ham. Given a set of mails which contains both ham mails and spam mails, when this set of mails are passed as the input to the detector program the detector program classifies the each mail either as a spam or ham based on the algorithm (set of conditions). In phase 1, a rule based spam detector was developed, which classifies the mail either as spam or non-spam based on certain rules. A well-know 250 blacklisted words which frequently occur in spam mails are collected and stored in a bag-of-words. The detector program check if the email contains any of these black-listed word in it. If the mail has black-listed word(s), then the mail is marked as spam mail, else the mail is marked as ham mail. In phase 2, a spam detector program was developed which classifies a mail as a spam or a ham not just based on the occurrence of the black-listed words. Because the black-listed words can also occur in ham mails. To overcome this drawback, a new algorithm which follows Naïve Bayes classification method was implemented. Initially the detector is trained with a sample set of mails which contains both spam and ham mails. Also each time when the word repeats the number of occurrence of that word is incremented by count one. Finally, two list (one for spam words and other for ham words) along with occurrence count of the words are stored in two hash list (dictionary).

NETWORK-MANAGEMENT-USING-SDN-and-NFV

This project builds on the idea of emerging concept of Software Defined Networking (SDN) and Network Function Virtualization (NFV). The basic notion of SDN is separation of Data and Control plane whereas the NFV is to interchange dedicated network appliances with software running on servers. The main concept behind project V-NM is to replace the traditional network management by modern approach using the above mentioned techniques. OpenFlow protocol and OpenStack framework are used in the implementation of this project, which allows to manage SDN and NFV with great flexibility

Unix_MemoryManagementSystem

It is proposed to develop a memory management system that would allocate and manage pools of memory on the heap. Typical operations that the memory management system would support include allocation/deallocation of memory pools, allocation/deallocation of memory chunks, and viewing the status and contents of pools.

WifiStarving

This project involves design and simulation of various flows in a single-cell Wifi network. This project aims to study the cause and effects of various Wifi issues such as “Wifi-Starving” and “Hidden Terminal” problem. NS3 is used to design the network topology and for obtaining the simulation results. Basic starvation and Co-operative starvation of Wifi nodes were measured. Both symmetric and asymmetric flows in the Wifi node scenarios were simulated and various parameters such as back-off interval, average collision probability using Bianchi's Formula.

SIPSpeaker

In this project, we have developed a SIP Server/Speaker (SIP UA) using JAVA, which can be considered as a robot which automatically answers to the incoming SIP calls. The SIP speaker will waits for the incoming call and answers when call is received from the Client. When the call is answered, a pre-configured voice message will be played out to the client and finally the call is terminated. A web server is also provided, where the administrator of the SIP Speaker can type a customized message through the web portal. This new message will be saved and played out to the SIP clients then on. Also the administrator can view, delete the existing message through the web page. For text to voice conversion, we used FreeTTS and “kevin16 voice“.

PacketBouncer

In this project, we have developed a packet bouncer, which can thought as an intermediate node/firewall which doesn’t allow the client inside the network to connect directly to the server. Bouncer listens and receives packet on fake IP address and on a specific incoming port. The client will send the packet to the bouncer on the specific IP and port. LibPcap is used to tap the received packet on the bouncer. The packet is validated on all fields on the IP header (Destination address, Checksum, TTL, Evil bit). If it was an ICMP packet then ICMP processing and validation is done, else if the packet contains TCP segment, then TCP checksum is validated. Finally TCP source and destination port and TCP checksum is calculated. Source and destination IP are modified and IP layer checksum is calculated and forwarded to Server. For the return journey of the packet, the similar processing is done expect the IP address field and ports number differs. If the packet is a FTP packet then control packet and data packet were handled in a different manner. Also the bouncer is designed to handle multiple clients at the same time. The project was developed using C language.

RUDP

In this project, we have developed a library for Reliable UDP, which enables the application using it to send multiple files to multiple destinations. The library inserts a RUDP header on the payload of normal UDP datagram by which a reliable transmission can be achieved. It does the flow control using Go-Back-N ARQ sliding window technique same as in TCP, but in contrast RUDP doesn’t provide congestion control, error recovery and other TCP capabilities. RUDP has its application in the scenarios where user do not want to add more overhead to the data (as like in TCP header of 20 bytes), but requires overall reliable data transfer.

Webmail

In this project, we have developed a webmail server (Http server + SMTP server) for sending emails using HTML and Java. Http server issue an e-mail form on client’s request. User can specify FROM and TO address, Subject, Email content, scheduled time to send the email, and optionally the receiver’s domain SMTP server address. If not mentioned an MX record lookup will be made to find the address of the receiver’s domain SMTP server. On submitting the form, local HTTP server receives the form, validates and forwards it to the local SMTP server which then tries establishes the connection with remote SMTP server. On successful handshakes, the local SMTP server forwards the email to the remote SMTP server. Also MIME support is also provided using “Quoted Printable” encoding and ISO-8859-15 character set is supported.

An-Efficient-Algorithm-For-Analyzing-Telephone-Signaling-In-International-Switching-System

This project deals with algorithm for monitoring the QoS and to detect failures in voice communication systems. Algorithm has two parts. First part is the analysis of data stored in Call Detail Records (CDR). For each call one CDR or ticket or traces is generated. These traces contain data related to the call describing the system elements involved, such as time and duration of the call, phone types and numbers, SS7 signaling, trunks, time slots, what happen to the call, etc. The tickets are generated in PSTN switches or over VoIP gateways, for the case of Internet Protocol Detail Record (IPDRs). These tickets can be used to monitor the QoS and to detect faults focusing in different aspects related to the call, such as technical, economic, or social. Second part is the calculation of Answer to Seizure Ratio (ASR), by which a communication network performance can be calculated. Our main goal is to analyze and produce the various important parameters from the CDR thereby helping the Network Performance Engineers to easily understand the performance of the network and to save their time.