Acoustic Localization of Mobile phones in Car for Driver Safety Applications

This project aims at improving the driver safety by appropriately allowing or denying calls to the driver’s mobile phone. The mobile phones equipped with microphones, calibrates its location within the car to determine if it is held by the driver or the passenger and accordingly enforces call policies. The technique that we propose makes use of human in-audible acoustic signals from the car’s speakers to calibrate the location of the mobile phone.

Vehicular Speed Estimation Using Received Signal Strength from Mobile Phones

This project focuses on estimating vehicular speeds with high accuracy at the base station using the mobile phones in vehicles without the explicit participation from the drivers. The work is founded on the principles that RSS from Mobile phones on the GSM network are stable over time and variable over space. We apply classic dynamic programming techniques to estimate vehicular speeds with very high accuracy This technique can be more robust to small scale fading and can produce more accurate speed estimation compared to the traditional technique of localizing phones over time since we are now looking at a continuous time series instead of discrete signal strength readings.

RIDE:Reliable Identity-spoof Detection and Elimination

Wireless Networks are vulnerable to a variety of identity spoof attacks where an attacker can forge the MAC address of his wireless device to assume the identity of a legitimate user. In this work, we propose mechanisms to detect MAC address spoofing using a combination of tamper proof metrics that rely on physical layer parameters such as RSSI and MAC layer information such as the IEEE 802.11 MAC Sequence number.

Understanding the Similarities in RSSI for Co-Moving wireless transmitters.

With the proliferation of 802.11b/g Wireless Devices, it is very common to have more than one wireless transmitter in close proximity. DECODE detects such transmitters that move together (Co-moving transmitters) by identifying correlations in communication signal strength due to shadow fading. It requires no changes in or cooperation from the tracked devices other than sporadic transmission of packets and can be detected from just a single receiver. Co-movement information can find use in applications ranging from inventory tracking, to social network sensing, and to optimizing mobile device localization.

Empirical Evaluation of the Limits on Localization Using Signal Strength

Wi-Fi Localization has reached a point where the accuracy limitations have to be overcome to realize its real potential in several of the ubiquitous computing applications. However, it is not clear what factors innately limit the localization accuracies to greater than 1ft as reported by several of the recent research. To understand this better, we experimentally analyze the different limiting factors in the presence of a high density wireless AP deployment. We also show that the experimentally achievable lower bounds are better than the Cramer Rao Lower Bound(CRLB) invalidating the typical assumptions behind CRLB in the real experimental environment.

Distinguising Environmental Mobility from Transmitter Mobility for Improving Localization stability

While several previous studies have characterized mobility based on RSS variance, it needs to be understood that RSS could also vary a lot if the objects in the environment cut the Line of sight path between the transmitter and receiver. In this study, we understand the differences in RSS variance induced by object mobility as compared to the RSS variance induced by environmental mobility and propose this as a framework to improve the stability of localization.

Optimizing Broadcast Load in Mesh Networks using Dual Association

This project aimed at optimizing the broadcast traffic load in a mesh network. Traditionally, association is based on the strongest signal strength. In this project, we examine the concept of multi-association, where the client chooses the access point for broadcast traffic and unicast traffic independently by exploiting multiple coverage that are typical in mesh networks. We proposed a novel metric called normalized-cost that is advertised in the beacons from APs. We showed that greedily associating with the AP advertising the least cost can reduce the broadcast traffic load significantly in the network. We also evaluated the association algorithm using real-experiments with the sensor nodes from the Kansei testbed http://ceti.cse.ohio-state.edu/kansei/.