NJIT's REU Site in Computer Networking and Security

Our 2007 program featured several research projects, as well as a short course on computer networking and security and several workshops thorughout the summer research experience. The program culminated with the NJIT Summer Research Symposium at which students participating in this and other summer research programs presented posters describing their projects. Below is a list of the projects, participating students and mentors, and links to the poster presentations.

Click on a topic to view the introduction to the project

1. Underwater Communication in Shallow Waters details

Mentor: Prof. Ali Abdi

Research Scholar: Janyll Perez, New Jersey Institute of Technology poster

Improvement for underwater communication is very important, not only to find out more about the mysteries under the seas, but also for safety. For example, the US NAVY uses Unmanned Undersea Vehicles (UUVs) during war to detect mines in shallow waters, which saves lives. Underwater communications had been successful for many years in deep water, but this is not the same in shallow waters. This is due to two major causes: multi-path propagation and non-Gaussian noise. Multi-path propagation increases the inter symbol interference (ISI) this causes frequency fading, limiting the communication. The non Gaussian noise is a disadvantage due to the fact that most wireless communication tools are designed for Gaussian noise. In this research we seek to understand underwater communication and to see how it changes by multiple receivers and different sea bottoms.

2. Turning an Office Into an iSpace details

Mentor: Prof. Mengchu Zhou

Research Scholar: Justen Walker, New Jersey Institute of Technology poster

An Intelligent space is an environment that can gather information about environmental changes and activities which take place therein. In an office environment, the former includes temperature, noise-level, humidity, and lighting conditions. The latter mainly includes activities of people working in front of some personal computer devices. An elementary approach for gathering information about the nature of activities performed on a personal computer is described and analyzed. Although this method is simplistic, it can be implemented as a real-time system. It is able to cue other devices in order to arouse or encourage more activity performed on the PC, thereby promoting the productivity.

3. Measurement of Memory Needs for Fast IP Routing details

Mentor: Prof. Roberto Rojas-Cessa

Research Scholar: Courtney Enoex, Souther Illinois University poster

Demands in the marketplace are leading to higher bandwidth network designs. With links that can carry over 30 million 40-byte packets a second, the need for efficient algorithms is clear. This research seeks to address the challenges facing the development of the perfect routing algorithm not only in terms of speed, but also in memory and updating capacity. These qualifiers are examined in the contexts of both routing table and packet classification algorithms. While no algorithm has yet combined the ideal characteristics for either routing function, there are promising candidates. This research examines a routing search tree proposed by Rojas-Cessa, et. al. This algorithm utilizes expanded prefix and parallel search techniques that will find a packet’s “next hop” router in under two memory accesses. Although intense study of the algorithm is necessary, this work demonstrates the usefulness of speed, memory, and updating measurements to reveal the promising qualities of the algorithm. The research uses the same measurements to discover any positive quality of the Fat Inverted Segment tree—a range search tree used in packet classification.

4. Novel Sub-wavelength Screens for Sensor Applications details

Mentor: Prof. Haim Grebel

Research Scholar: Jason Lin, Washington University in St. Louis poster

Simulations were conducted involving metallodielectric screen filters with various patterns for suitability for broadband filter design. Two main patterns were examined: crosses and cylindrical round holes. This research found that the cross patter produced a moderate slope curve that is useful for filtering a broad band of frequencies, while round, cylindrical openings resulted in a steeper slope curve that is more useful for narrow band frequency filtration. For the cylindrical holes, screens with thinner filter thickness resulted in a larger transmittance bandwidth.

5. A Framework for Analysis and Prototyping of IDS Alert Fusion Using Extensible Petri Nets details

Mentor: Prof. Edwin Hou

Research Scholars: Joseph Simons, Harvey Mudd College, and Luis Zaman, Xavier University poster

Intrusion Detection Systems (IDS) are notorious for generating a flood of alerts. Several methods for alert aggregation and correlation aim to alleviate this issue, however, there has been little work done to facilitate the development of new aggregation and correlation methods. Our approach uses the extensibility and simplicity of Petri Nets as a framework for modeling intrusion detection systems. We propose a novel tool for Rapid Analysis and Prototyping of Intrusion Detection Systems (RAPIDS) using extensible Petri Nets and a graphical user interface.

6. Characterizing the Effects of Low-Rate DoS Attacks on Streaming Media Traffic details

Mentor: Prof. Nirwan Ansari

Research Scholar: Kimberly Glaspey, Gonzaga University poster

Much research has been done to evaluate the effects of low-rate DoS attacks on TCP traffic in the Internet. However, very few studies have evaluated the effects of low rate DoS attacks on UDP traffic in the Internet, such as streaming media traffic. Advances in compression technology and the completion of the last-mile broadband networks have made video streaming a staple application in the Internet. For this reason, it is important to investigate the effects of the low rate DoS attack on the quality of received video. We used the video evaluation toolset EvalVid with the network simulator ns-2 for the extraction of video performance metrics such as PSNR, loss, and delay rates. EvalVid generates trace files which are put through an ns-2 simulation. This creates send and receive trace files which EvalVid then uses to make calculations and produce a potentially corrupted video. These results allow us not only to evaluate network statistics such as loss rates, but also allow for evaluation of the user-perceived video quality through Peak Signal-to-Noise (PSNR) and Mean Opinion Score (MOS) data. Our work can be beneficial to network defense engineers in building robust defense systems against these new breed of DoS attacks and will also greatly improve our understanding of these attacks.

7. Code Sequences for DS-CDMA Communications details

Mentor: Prof. Ali Akansu

Research Scholar: Elizabeth Diedrich, St. Mary's College poster

Code Division Multiple Access (CDMA) uses pseudo-random codes in order to allow for multiple users on a single channel. Previous technology required the users to divide their use by time (Time Division Multiple Access: TDMA) or by frequency (Frequency Division Multiple Access: FDMA). In CDMA, pseudo-random codes work as identifiers for the users of channels. The type of code used in CDMA transmissions is important because it determines the number of users per channel, the security of the channel, and the method of transactions that can be completed on that channel. The original codes used were Walsh codes. My research focused on producing new code sequences that would be compatible with the previous technology. I tested my codes in situations with 2-user in synchronous and asynchronous communications with AWGN (Additive White Gaussian Noise) and found that the codes performed almost equally in the synchronous case and better in the asynchronous case.

8. Statistical Approach for Location Estimation in Wireless Networks details

Mentor: Prof. Quentin Jones

Research Scholar: Eric Rebeiz, University of Massachusetts at Amherst poster

Nowadays, with the extended growth of mobile computing devices such as cell phones and computer along with local area wireless networks, network providers have started to add value to the services offered to the users such as users’ location awareness systems. Thus, tracking a person in indoor wireless network has become an increasingly important issue. This project introduces a new statistical method for location estimation in a wireless network and studies the variables that may affect the signal strength in different situations. Although this idea of Radio Frequency based technique for user location estimation is not very new, our algorithm is the first one to that combines easy training sets with considerably high accuracy of estimation. We have tested our system on 802.11b wireless network in the GITC building and our algorithm paws able to achieve an accuracy of location estimation within one cell of the actual location with a probability of 92%.

9. Practical Packet Classification Schemes for Extended Network Services details

Mentor: Prof. Roberto Rojas-Cessa

Research Scholar: Brian D'Alessandro, New Jersey Institute of Technology poster

High speed and efficient packet classification is a core issue in modern router hardware design. Many different algorithms exist to match an incoming packet to a list of firewall rules, however, they do so with varying performance metrics. This project studied packet classification using the Cross-Producting scheme, TCAMs, and a mixed solution of the two. Performance was evaluated by the number of memory accesses required to match a packet, as well as the amount of memory the solution required.

2007 Program