Researcher profile: Tashniba Kaiser

Tashniba Kaiser is a master’s student and graduate research assistant at the University of Manitoba’s department of electrical and computer engineering. Kaiser, a native of Dhaka, Bangladesh, completed her bachelor’s degree in engineering at the Military Institute of Science and Technology in Bangladesh.

Kaiser worked at Airtel Bangladesh Ltd, a leading mobile phone company for four years where she gathered experience in wireless networks. She went on to work in the data network and security field department of IBM Bangladesh for eight months as a network engineer. After much training, she now works with Dr. Ken Ferens at the U of M, studying “Wireless Device Localization Using Artificial Neural Network.”

“Everyone is entitled to their privacy,” Kaiser told the Gradzette. “Organizations need to know when an intruder is attempting to gain access to their network or protected data and to quickly take corrective measure to thwart and deny access when required.”

Wireless theft has become a big issue in recent years. A lot of network owners and users don’t see the depth of danger ahead with increasing reliance on computer networks as a means for transferring personal information and data.

Kaiser’s study proposes a method to track and locate strange wireless devices or signals within the premises of a given organization. Security protocols are set up to prompt the identification and authentication of guest users or devices trying to gain access to a protected network and to deny access when necessary.

Kaiser is using artificial neural network (ANN) to localize the wireless device. This algorithm is trained to learn the environmental impairments of an organization’s premises and investigate the position of a wireless transmitter for implementing the required security policy.

“Artificial neural networks are a bunch of computers organized like our brain,” Kaiser explained. “By providing the ANN with various contaminated samples of the signal strengths for each known location in the protected area, the ANN will learn to associate the impaired samples to the originating position. This is how we can locate a wireless device.”

Training an artificial neural network requires a neural network toolbox in Matlab. Three wireless receivers were placed in the protected premise (wireless infrastructure) to locate the position of the wireless client. Triplet sets of signals were received from these three receivers.

Received signal strength indicator (RSSI) is a measurement of the power present in a received or an incoming signal. The RSSI circuit is designed to pick RF signals and generate an output equivalent to the signal strength. The ability of the receiver to pick the weakest of signals is referred to as receiver sensitivity.

ANN was used to learn the characteristic features of the signal strength captured from targeted device via RSSI. The training data set is fed initially to the ANN with a specific learning rate and the errors are adjusted, Kaiser explained. With this, it automatically detects a wireless-device within the proximity of its premises and based on certain calculations, it will either grant or deny access to the protected wireless network.

According to Kaiser, “we rely on sensing RSSI readings to identify the actual signal that has been received from the receivers. Unfortunately, these measurements are often contaminated with channel errors, which include time varying impairments and environmental impairments.”

Kaiser’s major problem is interference. RSSI suffers from interference from other deterministic narrowband wireless technologies, such as Wi-Fi, Bluetooth, ZigBee, cordless phones, microwave ovens. Other problems encountered include fading, shadowing, multipath, non-line-of-sight propagations, reflections, and attenuation from obstacles in the environment. Despite these predicaments Kaiser’s work has seen the daylight to an acceptable extent.

Her gratitude and appreciation goes out to her supervisor who has always showered her with encouragement and moral support. “I am highly indebted to him,” Kaiser told the Gradzette.

This article was originally published in the Gradzette.