Students' Research Presented at the WiSec International Conference-Silicon Valley Campus - Carnegie Mellon University

Wednesday, May 8, 2013

Students' Research Presented at the WiSec International Conference

A technical paper by a Carnegie Mellon team was accepted by The Sixth ACM Conference on Security and Privacy in Wireless and Mobile Networks (WiSec). INI student Shrikant Adhikarla (MS23), ECE student Min Suk Kang along with Silicon Valley faculty member Patrick Tague, assistant research professor at the INI and CyLab, authored the paper, titled "Selfish Manipulation of Cooperative Cellular Communications via Channel Fabrication." WiSec is a major industry conference that took place April 17-19, 2013, in Budapest, Hungary.

Adhikarla, a second-year graduate student in the Master of Science in Information Technology - Information Security program, reported that the team's research began as coursework for the Wireless Network Security course during spring semester 2012. Adhikarla, Kang and Dr. Tague proposed a channel fabrication attack by user equipment on cooperative cellular systems.

"Basically, the LTE cellular systems have recently proposed cooperative cellular networks, wherein geographically-separated multiple base stations cooperate on transmission in order to improve the signal-to-interference-plus-noise-ratio (SINR) at cell-edge region for users of mobile devices. Although it is the service provider that makes the clustering decision for the user, deciding which base stations should cooperate to improve the signal heavily relies on the trust in the channel values reported by user equipment (mobile phones). In this paper, we propose a new attack against the cooperative cellular networks," said Adhikarla.

"It's necessary for the service providers to be aware of this weakness in the system and have proper counter-measures in place to avoid such attacks," explained Adhikarla. "In the paper, we discuss an anomaly detection mechanism for the service provider to detect the attack with approximately 90% of accuracy."

"Shrikant and Min Suk's research demonstrates the potential security implications of our rapidly advancing telecommunications infrastructure," said Dr. Tague. "Telecom service providers should really consider such security issues in their system design, especially if they want to provide the level of service that customers are expecting."

The paper is available online at the following link: http://dl.acm.org/citation.cfm?id=2462105

Full Abstract: In today's cellular networks, user equipment (UE) have suffered from low spectral efficiency at cell-edge region due to high interference from adjacent base stations (BSs), which share the same spectral radio resources. In the recently proposed cooperative cellular networks, geographically separated multiple BSs cooperate on transmission in order to improve the UE's signal-to-interference-plus-noise-ratio (SINR) at cell-edge region. The service provider of the system dynamically assigns the cluster of BSs to achieve higher SINR for the UE while optimizing the use of system radio resources. Although it is the service provider that makes the clustering decision for the UE, the service provider relies on the UE's input to the decision; i.e., the channel states
from the adjacent BSs to the UE. In essence, the operation of the cooperative cellular networks heavily relies on the trust in the UEs. In this paper, we propose a new selfish attack against the cooperative cellular networks; an adversary reprograms her UE to report fabricated channel information to cause the service provider to make a decision that benefits the adversary while wasting its system resources. We evaluate the proposed attack in a cooperative cellular network having various performance goals on the simulation-based experiments and show that the adversary can trick the service provider into expending 3.7 times more radio resources for the adversary and, accordingly, the adversary achieves up to 16 dB SINR gain. Finally, we propose a threshold-based countermeasure for the service provider to detect the attack with approximately 90% of accuracy.