You are on page 1of 7

PROJECT LIST Department of EE Fall 2013 SUPERVISOR TITLE

Dr. Junaid Qadir

ABSTRACT

PRE-REQUISITES

CHALLENGES

Formal verification in SDNs Anomalous behavior in computer networks has been traditionally very difficult to troubleshoot due to the sheer complexity of such networks. Traditionally, networks are constructed from devices that are managed and configured on a device-by-device basis through vendor-specific proprietary command-line-interfaces. Since the network configuration is on a device-by-device basis, its difficult to implement network wide policies consistently. It has been shown that most errors are results of human misconfiguration (some reports put this at up to 80% of all network outages) . Such errors are exacerbated by the traditional network architecture of coupled data, control, and management plane which do not allow management (or programming) of the network as a system. Due to the complexity of traditional networking, and diversity of heterogeneous control-plane implementation on a per-device basis, networks allowed very little formal reasoning and verification about the correctness of network behavior. The SDN architecture which proposes decoupling of data and control planes of individual devices allows network managers to program the network as a system allowing consistent enforcement of policies. This architecture proposes centralized implementation of control plane (as a distributed system) which can control the data planes of network devices remotely through a standards-based interface. This can significantly simplify network configuration, management and operations. This architecture also allows us to reason about network behavior in a manner that was previously infeasible. The following tasks that could not be done previously in an effective manner have now become feasible: 1) Formal verification about the correctness of a software control-plane module 2) Formal verification about the correctness/ consistency of planned network policy 3) Reasoning about network behavior The SDN architecture which proposes decoupling of data and control planes of individual devices allows network managers to program the network as a system allowing consistent enforcement of policies. This architecture proposes centralized implementation of control plane (as a distributed system) which can control the data planes of network devices remotely through a standards-based interface. This can significantly simplify network configuration, management and operations.

Good knowledge of networking and To be decided programming SE students who have taken a course on Formal Methods as part of their course work would be preferred due to their subject knowledge. Students from other streams can also apply if they are interested in this exciting field. Good work ethics which include an ability to work hard to grapple with problems no one has solved yet. Ability to work independently and also in a group.

Dr. Junaid Qadir

SDN with data mining/ big data

Prerequisites: To be decided Good knowledge of networking and programming Basic knowledge of machine learning and artificial intelligence Good work ethics which include an This architecture also allows us to utilize the troves of data that is typically available to network ability to work hard to grapple with operators about the usage pattern of the network (in terms of application usage, network problems no one has solved yet. utilization [both temporal and spatial], user behavior, etc.) This allows an exciting opportunity for Ability to work independently and mining meaningful information about traffic/ flow logs which can allow customizing and also in a group. adapting network behavior and offering new services. Network operators can use data mining/ big data techniques to: a. Utilize their network more efficiently: operational costs can be driven down by effective traffic engineering in the network to ensure that network resources are efficiently utilized. b. Offer new services: network data has information along many dimensions including timerelated information. Network operators can open up new innovative services that exploit the temporal patterns of usage while also ensuring more efficient utilization of network resources. c. Detect anomalous behavior: flow-level data can be mined to infer about network security attacks and anomalous behavior. Big data techniques which allow processing of huge amounts of data on clusters of computers can be useful for processing the huge amounts of data that will be available. These techniques will be complemented by machine-learning techniques that allow learning from data to get useful

actionable information. Dr. Junaid Qadir Improved network management for SDNs The SDN architecture which proposes decoupling of data and control planes of individual devices allows network managers to program the network as a system allowing consistent enforcement of policies. This architecture proposes centralized implementation of control plane (as a distributed system) which can control the data planes of network devices remotely through a standards-based interface. This can significantly simplify network configuration, management and operations. Traditionally network management has depended mostly on simple protocols such as SNMP, Syslog, flow monitoring tools such Netflow and troubleshooting tools such as ping, traceroute, etc. Network configuration, on the other hand, has been large manual with management tools such as SNMP allowing very inflexible support . Unfortunately, the reliance on human configuration and management has also led to many unnecessary network outages---it has been shown that most errors are results of human misconfiguration (some reports put this at up to 80% of all network outages) . Most of the current state-of-the-art network management tools (e.g., HP OpenView) are essentially glorified GUI frontends for SNMP (and similar protocols). However, the new architectural paradigm of SDN, and the availability of new management protocols such as Netconf, opens up a new world of possibilities for the functionality that can be made available at a Network Management System (NMS). The new NMS should allow seamless monitoring, management, reporting, visualization, troubleshooting, and configuration. We will focus our research on building a system that takes us further towards this vision of a highly functional NMS with these features. The SDN architecture which proposes decoupling of data and control planes of individual devices allows network managers to program the network as a system allowing consistent enforcement of policies. This architecture proposes centralized implementation of control plane (as a distributed system) which can control the data planes of network devices remotely through a standards-based interface. This can significantly simplify network configuration, management and operations. This architecture also allows us to utilize the troves of data that is typically available to network operators about the usage pattern of the network (especially in terms of network utilization [both temporal and spatial]). This allows an exciting opportunity to utilize AI and optimization techniques to automatically adapt routing behavior in network to allow load balancing and effective utilization of resources. Implementation of such adaptive routing is made feasible by the architectural support of SDN which allows computation of routes through intelligent algorithms based on global network information at a central location which then can install appropriate flows at different devices in the network. This project will be ideal for students who are interested in both wireless networking and machine-learning. Our focus will be on proposing machine-learning based routing protocols for cognitive radio networks. Cognitive radio networks are networks of nodes equipped with cognitive radios that can adapt to the network conditions to optimize on performance goals. Prerequisites: To be decided Good knowledge of networking and programming Basic knowledge of machine learning and artificial intelligence Good work ethics which include an ability to work hard to grapple with problems no one has solved yet. Ability to work independently and also in a group.

Dr. Junaid Qadir

Adaptive routing in SDNs

Prerequisites: To be decided Good knowledge of networking and programming Basic knowledge of machine learning and artificial intelligence Good work ethics which include an ability to work hard to grapple with problems no one has solved yet. Ability to work independently and also in a group.

Dr. Junaid Qadir

Machine-learning based algorithms for cognitive radio networks

Dr. Junaid Qadir

Intelligent transport systems Intelligent transport systems (ITS) are advanced applications which aim to provide innovative (ITS) services relating to different modes of transport and traffic management and enable various users to be better informed and make safer, more coordinated, and 'smarter' use of transport networks.

Prerequisites: To be decided Good knowledge of networking and programming Basic knowledge of machine learning and artificial intelligence Good work ethics which include an ability to work hard to grapple with problems no one has solved yet. Ability to work independently and also in a group. Prerequisites: To be decided Good knowledge of networking and programming Basic knowledge of machine learning and artificial intelligence Good work ethics which include an ability to work hard to grapple with problems no one has solved yet.

Dr. Junaid Qadir

Learning analytics

Learning analytics (for learning management systems) (2 students) With increasing amounts of data about the learning habits and history of students being online in LMS, there is a great opportunity for using machine-learning techniques for mining useful knowledge about the students. This project will focus on efforts in this domain.

Dr. Junaid Qadir

Living analytics

Living analytics (social science with mobile networking) (2 students) For a sample look at the research domain, have a look at the work done by Living Analytics Research Center (http://centres.smu.edu.sg/larc/)

Dr. Junaid Qadir

Natural language processing Natural language processing Working with Arabic language and Islamic studies (2 students) Example research being done in this domain http://www.comp.leeds.ac.uk/arabic/dukes11pgr.ppt

Dr. Osman Hasan

Analysis of Power Distribution management system algorithms in Smart Grids

Voltage ratings of a Smart Grid must be curtailed within some given limits. Moreover, the fault restoration problem of smart grids is a major challenge. There are many random factors, such as load, power production and network configuration variations due to DG penetration, that affect these limits and thus a rigorous analysis to ensure that the voltage limit bounds are not violated and fault restoration is always possible is required. The main scope of this project is to conduct these analyses using PRISM, which is a Markov chain based analysis tool.

Dr. Osman Hasan

Dr. Osman Hasan

Wireless E-health module The project aims at developing a wearable belt with electrodes to record the Electrocardiography for Blood pressure and ECG (ECG) signals and blood pressure. The belt would also contain a microcontroller and a monitoring communication module to transmit the acquired data to a host server, maintained at a central facility via GPRS. The students would work on developing the complete embedded system and the associated web server, which would also provide the ability to track the patients wearing the belts using the GPS signals. Formal Verification of As the semiconductor industry moves towards on-chip many-core architectures, the optimal Distributed Thermal management of the available heat dissipation has become the foremost concern. The thermal Management Schemes for management deals with the problem of distributing the computation tasks among the cores to on-chip Many-core Systems keep the overall temperature of the chip under some given threshold. However, this problem is NP-hard and centralized schemes quickly become infeasible due to lack of scalability when entering the many-core era. Distributed thermal management schemes have shown promising results in achieving a near-optimal solution by breaking the complete problem into smaller ones, based on the principles of locality, and then exchanging state-information across the distributed boundaries via agent negotiations for decision making. The choice of tuning parameters for this negotiation has been identified as the most critical aspect in ensuring a stable system. In this project, we will use mathematical techniques and associated computer based tools to analyse

Ability to work independently and also in a group. Prerequisites: To be decided Good knowledge of networking and programming Basic knowledge of machine learning and artificial intelligence Good work ethics which include an ability to work hard to grapple with problems no one has solved yet. Ability to work independently and also in a group. Prerequisites: To be decided Good knowledge of networking and programming Basic knowledge of machine learning and artificial intelligence Good work ethics which include an ability to work hard to grapple with problems no one has solved yet. Ability to work independently and also in a group. Prerequisites: To be decided Good knowledge of programming Basic knowledge of machine learning and artificial intelligence Good work ethics which include an ability to work hard to grapple with problems no one has solved yet. Ability to work independently and also in a group. Understanding about algorithms and power distribution systems is sufficient for this project. The rest of the required skills, such as understanding of Smart Grid Systems and the PRISM model checker would be acquired during the course of the project. Microcontroller based design Sensor data acquisition

Markov Chain Theory (Probability Theory) Basic concepts about Multi-core ICs Microprocessor systems

these parameters and thus devise an optimal thermal management scheme. Habeel Ahmad Pollen Counter The project aims to develop a pollen counting device based on image processing techniques. The Pollen counter will pass the ambient air through a known cross sectional area. The Snap Shots of Pollen particles shall be taken using a high resolution camera for a given time interval corresponding to one meter cubic volume of air flow in front of the camera. The Pollen particles detected in each snap shot shall be graded according to their size/type and a count of each grade/type shall be recorded. The average count of various Pollen types count shall be displayed on an LCD Speech processing is the study of speech signals and it deals with the processing methods of these signals. The signals are usually processed in a digital form; therefore speech processing can be regarded as a special case of digital signal processing. The applications of speech processing include speech recognition, speaker recognition, speech coding, voice analysis, and speech synthesis and speech enhancement. Matlab Programming Image Processing Techniques/Algorithms Micro-Controller Programming/Interfacing Image Processing & Display Pollen Collector Unit

Dr. Muhammad Shahzad Younis

Sentiment Analysis from Speech Signal

Matlab, FPGA

While processing the speech for speaker identification or speech recognition it is becoming important to extract the sentiments/expressions of a speaker from the speech signal. Primary sentiments are happiness, sadness, anger, fear, surprise and disgust. Different features of speech like, pitch, energy, formants can be utilized to extract these sentiments from the speech signals.

Dr. Muhammad Shahzad Younis

Direction Finder

Dr. Muhammad Shahzad Younis

Dual Core RISC Architecture

Dr. Adeel Baig

IPv6 Monitoring Suite

Speaker identification is not sufficient in number of environments where it is critical to identify the position and direction of the speaker along with who is speaking e.g. in disaster management, automatic mic controlling mechanism, speech controlled cameras etc. An efficient direction finding algorithm is required which not only precisely detects the direction of the speaker but it has to be in a real time manner in order to support time critical applications. RISC stands reduced instruction set computer which is used for dedicated signal processing, control and image processing applications among others. Idea is to design a RISC architecture which incorporate two data paths each with its own data memory, single program memory and a single controller. This is will not only accelerate the program execution but also supports the concurrent multiple process execution to support the real time applications. Network and services monitoring has become more important due to their usage for our critical operations. It is important that we are able to monitor our solutions and have more control over our network and services. Network monitoring has been performed for IPv4 networks for a long time. In case of IPv6, there are many documents and guides explaining how to deploy IPv6 networks and services but there are very few that describe in detail how to monitor IPv6 network. This project requires the students to develop IPv6 network, services and applications monitoring suite.

FPGA, Matlab, Verilog

Verilog, FPGA

Good understanding of IPv6 Good Understanding of SNMP Linux Programming Java/C++

Development of monitoring modules for: Web servers Applications Services Network Entities

Dr. Adeel Baig

WLAN optimization controller for Triple Play Services

Dr. Adeel Baig

Wireless Local Area Network (WLAN) technology is used to offer various data services to endusers over the last-mile. These data services include video, voice and data traffic and together are frequently referred to as the Triple-Play services. These services have different quality of service (QoS) and bandwidth demands and must be treated accordingly. WLAN optimization controller is proposed to provide the required QoS by dynamically adjusting the QoS parameters for different applications and users. Last year a module for static QoS allocation was implemented and tested. This project will be a significant extension to that project. WiFi Communicator for iOS Wireless Local Area Network (WLAN) or Wifi technology is used to offer complete communication solutions to the users. It offers incredible high speed and a long range. The potential of using Wifi technology in this way is promising as it can significantly reduce infrastructure and maintenance costs inherent in cabled architectures. Similar projects were developed for Android and Windows Phone last year.This project requires the students to develop a P2P communicator, for iPhone iOS, that uses the bandwidth of wireless access point and provides a complete communication solution (voice, chat, content, video). This solution should

Linux/Kernel Programming Very good understanding of WiFi Aptitude for research C/C++

Strong background in iOS development Good understanding of VOIP/SIP protocols GUI development

Development of QoS Algorithm Mapping proposed algo to WiFi Modifying Linux Kernel to support the also Implemntation/Testing/GUI P2P node discovery and address assignments App for Data/Voice/Video iOS development challenges

Dr. Adeel Baig

Dr. Muhammad Shahzad Younis

Dr. M. Usman Ilyas Dr. M. Usman Ilyas Dr. M. Usman Ilyas

work without the need of a separate sever. The clients coordinate and communicate in a distributed Adhoc manner (from address assignment to complete data transfer). QoS in IPv6 Networks using QoS (Quality of Service) provisioning is an indispensable part of communication networks. SDN Approach Software Defined Networking (SDN) introduces new features that enable network administrators to have more control over their network deployments. IPv6 is the future of internet. IPv6 provides a large address space and greater flexibility for introducing new options in the future internet. The latest SDN specification provides only a limited and not well-defined QoS framework. In this project students will develop an integrated QoS framework for SDN. As IPv6 networks supporting direct flow handling (by the Flow Label in the protocol header), with the combination of SDN based forwarding control, this can be an efficient solution to manage future large scale networks. Hardware architecture for Image compression is used in number of applications where transmission bandwidth is not lossless image compression sufficient to support real time data generation. In transmission of medical signal/images/videos researchers employs loos-less compression due to the critical nature of data where details are mandatory to be preserved for correct and reliable diagnosis of diseases. Real time applications demands a real time algorithm for data compression. Scope of the project is to design and implement a real time image processing architecture to compress the real time generated medical image. A Passive Crowdsourcing Monitor road conditions by crowdsourcing data collected by many Android smartphone users. App to Monitor and Map Road Conditions Identifying Leaders and identify influencers in an online social network. This work will be an extension of prior work Followers in Social donw by myself and published recently. Networks Physical Activity Detection Detect the physical activity the carrier of a cellphone is involved in at all times. Activities will Using include Walking running, driving, cycling, climbing stairs, resting. Work was done for Android earlier and needs to be redone for iOS. Web development This project would see through an implementation of a software defined exchange using principals of Software Defined Networking. A team of 2 students with BICSE/EE (Networks) background should be able to implement the project. Kindly discuss with me for specific details. The project team would work with ICTRDF funded project team (NUST, Bitsym and LUMS) in order to implement mine and miner monitoring suite. The work involves working both with devices (micro-controllers, transceivers, firmware design, communication and networking protocols) and front end application development in Java. This work would involve development and deployment of an internet censorship detection tool in Pakistan and would be deployed on devices deployed city-wide.

SDN foundation Network Programming C/C++ Python Research Aptitude

Developing modules for QoS Integration with SDN controllers System Implementation

Matlab, Verilog

Understanding of lossless compression algorithm, transformation the algorithm into hardware architecture

Android app development, signal processing, probability, machine learning Linear Algebra, good math skills, probability, graph theory, Python, Matlab Web development, iOS app development, simple machine learning / pattern classification concepts Networking Programming (Java/C/C++/Python) Firmware design (C/C++) Strong in Java application development Strong in communication and networking concepts Networking Programming in C/C++ and Java An implementation of a functional exchange using SDNs.

Dr. Saad Qaisar

Software Defined Exchange

Dr. Saad Qaisar

Mine Monitoring Application Suite

Dr. Saad Qaisar

Internet Censorship Detection Tool

Students would work with a team of researchers to develop internet censorship detection tool, port it on network devices, integrate them with controllers, measure censorship in realtime and document their results in form of a research article.

MUHAMMAD JAMEEL NAWAZ MALIK

A Self driving Toy Car

MUHAMMAD JAMEEL NAWAZ MALIK

FPGA Based Direct Digital Synthesizer (DDS) using CORDIC Algorithm

A self driving car takes in inputs for a final destination and drives itself from its initial location to the final destination. The idea for this project requires an LCD to display the location that the car is already at and allows a user to input coordinates for where he or she would like the car to drive itself to. By using a GPS receiver, the car can track its initial location and use the compass to direct where it should go, constantly updating its GPS location. This project aims to provide a Design and Implementation of CORDIC algorithm for Direct Digital Synthesizer to achieve a wide range of frequencies for applications like Software Defined Radios. COordinate Rotation DIgital Computer (CORDIC) algorithm is an interesting technique for phase to sine amplitude conversion. The CORDIC algorithm in this design is to utilize

1- Hands on experience on microcontrollers 2- Embedded C coding

1- Digital System Design Techniques 2- Verilog HDL

dynamic transformation rather than ROM static addressing. CORDIC algorithm provides fast and area efficient computations of sine and cosine functions without using ROM LUTs. Dr. Farooq Ahmad RFID Design (Active and Radio Frequency Identification (RFID) systems use radio frequency to identify, locate and track Tahir Passive) and Antenna people,assets and animals.Passive RFID systems are composed of three components a reader Circuit Design for RFID (interrogator), passive tag and host computer. The tag is composed of an antenna coil and a Applications silicon chip that includes basic modulation circuitry and nonvolatile memory. The tag is energized by a time-varying electromagnetic radio frequency (RF) wave that is transmitted by the reader. This RF signal is called a carrier signal. When the RF field passes through an antenna coil, there is an AC voltage generated across the coil. This voltage is rectified to result in DC voltage for the device operation. The device becomes functional when the DC voltage reaches a certain level. The information stored in the device is transferred to the reader by reflecting, or loading, the readers carrier. This is often called backscattering. By d etecting the backscattering signal, the information stored in the device can be fully identified. Dr. Farooq Ahmad To design and implement a Are you interested in building and testing your own imaging radar system? It includes the design, Tahir small radar system for range fabrication, and test of a laptop-based radar sensor capable of measuring Doppler, range, and localization of moving forming synthetic aperture radar (SAR) images. You do not have to be a radar engineer but it targets helps if you are interested in any of the following; electronics, amateur radio, physics, or electromagnetics. It is recommended that you have some familiarity with MATLAB. Teams of three students will receive a radar kit and will attend a total of 5 sessions spanning topics from the fundamentals of radar to SAR imaging. Experiments will be performed each week as the radar kit is implemented. You will bring your radar kit into the field and perform additional experiments such as measuring the speed of passing cars or plotting the range of moving targets. A final SAR imaging contest will test your ability to form a SAR image of a target scene of your choice from around campus; the most detailed and most creative image wins. Dr. Farooq Ahmad Mobile Phone Signal The rapid proliferation of cell phones at the beginning of the 21st century to near ubiquitous Tahir Jammers GSM/DCS, 20status eventually raised problems, such as their potential use to invade privacy or contribute to 520MHz, 2-3GHz (Noise academic cheating. They were originally developed for law enforcement and the military to generation techniques and interrupt communications by criminals and terrorists. Some were also designed to foil the use of amplification) certain remotely detonated explosives. The civilian applications were apparent.As with other radio jamming, cell phone jammers block cell phone use by sending out radio waves along the same frequencies that cellular phones use. This causes enough interference with the communication between cell phones and towers to render the phones unusable. Dr. Syed Ali Design and Development of Cloud storage is increasingly becoming popular due to the immense ease of use it provides. Haider Adaptive Cloud Storage According to certain estimates, in a few years most users will partially switch to cloud storage. Model With a host of applications enabling smooth transition of data from consumer end to the server end and back, an increase in network traffic is observed. Increase in traffic translates to congestion in the networks and increased delays. Against the same backdrop, this project aims at developing an adaptive cloud storage model that can efficiently utilize network (and datacenter) resources in the wake of increasing network traffic. Dr. Syed Ali Analysis of Cloud Storage A large number of cloud storage service providers have emerged in the recent past. The services Haider Services provided by these providers vary greatly in terms of synchronization latency, storage location, redundancy and security. In this project, we will carry out a comparative analysis of cloud storage providers exposing key parameters of efficient cloud storage.

Ability to work hard Ability to work independently Be self-motivated Good understanding of Electromagentic Field Theory and Antennas , Appropriate knowledge of Eelctronics/DSSP. Programming C/Matlab

Ability to work hard Be self-motivated

Ability to work hard Be self-motivated Good understanding of Electromagentic Field Theory, Mobile communication and Antenna Theory.

Computer Networks C/C++ or Python Wireshark Interest in the cloud Basic Probability Working knowledge of Virtual Machines Computer Networks C/C++ or Python Wireshark Interest in the cloud Basic Probability Working knowledge of Virtual Machines Students should have good skills in working on various hardware platforms. Also, he/she should be well versed in communication technologies and networks. 1. Design of a cost effective SPO2 sensor board. 2. Communicate the data over short distances using the best possible communication technology 3. Detection of breaches and raising of alarms.

Dr Adnan Khalid

Design and Implementation of SPO2 Sensor Board

The project involves designing a sensor board to extract SPO2 readings of patients and transmitting them over short distances to a relay node. SPO2 or Oxygen Saturation levels are one of the vital signs that need to be measured periodically and once the level exceeds a defined threshold an alarm needs to be generated and sent over to a gateway device.

You might also like