For more information, visit the Laboratory for Advanced Research in Information Assurance and Security (ARIAS.
Seeking a Secure Cognitive Network
The looming spectrum shortage created by the boom in wireless applications is gaining attention nationwide.
General news media covered the recent Federal Communications Commission (FCC) auction. Google has submitted a proposal to allow wireless broadband in the TV white spaces &mdash those unused areas in the spectrum. Meanwhile, engineers are racing to perfect cognitive radio technology that is expected to play a large role in more efficient spectrum sharing.
ECE's Jung-Min Park supports efforts to deal with the overcrowded unlicensed bands, but he wants to avoid security problems that plague other networks, such as the Internet. The best way to ensure security, he says, is to design for it from the very beginning.
Park has received a $430,000 National Science Foundation (NSF) CAREER Award to head the first major effort to explore non-conventional security issues in cognitive, spectrum-sharing networks. The five-year CAREER grants are NSF's most prestigious awards for junior faculty.
"It's like living in somebody's house at the same time as the owner &ndash without ever providing any clue that you are there."
"The spectrum shortage is hampering the development and deployment of new wireless applications and technologies," Park explains. "But, most of the spectrum has been already allocated for licensed use, and the FCC cannot allocate more bands for these emerging applications using the current regulatory paradigms. So, the FCC is considering opening up licensed bands &mdash such as the TV bands &mdash to unlicensed, secondary operations on a non-interference basis with the licensed users, who are typically called incumbents.
In such a scenario, the secondary operators would carry the burden of using the bands when available and ensuring they do not interfere with the licensed incumbents. The incumbents would not have to alter their operation or technology. "This is a hard problem," Park says. "It's like living in somebody's house at the same time as the owner &ndash without ever providing any clue that you are there."
One of the more promising technologies for these secondary operators is cognitive radios, which can sense and understand their environment and actively change their mode of operation, including frequency, modulation, and transmission power. Researchers &mdash including teams at Virginia Tech &mdash have developed working prototypes of cognitive radios and now are working on high performance networks.
"We expect that cognitive radio technology will be used for two-way communications in a wide range of applications, such as communication systems for tactical military forces and emergency responders."
The other major arena for the technology is the development of wireless access networks that can provide Internet services to rural areas, he notes. The world's first standard (IEEE 802.22) for wireless access networks based on cognitive radio technology is currently being developed for rural wireless access.
"If you think of security after everything is finished, it's too late."
The advantages of this technology, however can be offset by new security threats that have not been considered previously. "In a civilian cognitive radio network, the motive of a malicious user might be to simply cause mayhem to other users to receive notoriety. This would be the equivalent of computer hackers," he adds.
Malicious users also could try to extort money from providers who operate cognitive radio networks and services. In a military setting, an adversary could try to bring down a network or interfere with its communications to gain a tactical advantage.
"If you think of security after everything is finished, it's too late," Park says. "First developing a functioning network, then guaranteeing security doesn't work. The Internet is the perfect example, with its problems of spam, denial-of-service attacks, viruses, and Trojan horses. Let's break that paradigm. This time, let's think about security at the design stage."
Security is possible to implement from the beginning, according to Park. "These are policy-based radios. They all follow human rules written in software. They are not self intelligent."
Focus on Non-Conventional Threats
Park and his students are focusing on non-conventional threats. He defines conventional security threats as those elements that threaten the confidentiality, authentication, or integrity of the data within the network. "Conventional security threats can be countered using off-the-shelf cryptographic solutions, such as encryption and authentication schemes," he explains. "We use the term 'non-conventional security threats' to denote threats that cannot be thwarted with cryptosystems alone."
The team has identified three different security problems to explore and develop countermeasures against: incumbent emulation; Byzantine failures in cooperative spectrum sensing; and vulnerabilities in cognitive radio network self-coexistence.
Incumbent emulation would involve modifying the radio's software to change its emission characteristics to appear to other users as an incumbent. This possibility stems from the inherent programmability of cognitive radios and would undermine the spectrum utilization rules, ultimately reducing a network's throughput.
Byzantine failures would appear when incorrect sensing data is sent to the network from rogue terminals run by adversaries or faulty radios with malfunctioning software or hardware. "This problem involves determining how a network can make accurate estimations or predictions in spite of bad information. Can you come up with a scheme that maximizes the network's ability to make accurate predictions, even in a hostile environment where some of the nodes have turned bad?" Park queries.
The third major problem involves overlapping cognitive radio networks trying to operate in the same spectrum. He explains that this problem can only happen with multiple secondary user networks. Because cognitive radios are flexible and can easily switch bands &ndash or even support each other's traffic in a friendly manner &mdash there will be interference issues. "How will the radios change their modulation, power, and frequency in those situations?" The problem is called network self-coexistence. "It turns out that there are inherent weaknesses in the current 802.22 protocols for self coexistence. This is something that must be addressed before networks can safely operate in this mode," he says.
Probing a network before it exists introduces unique hurdles. Park foresees a big challenge in quantifying the security threats. "Finding and discussing these threats qualitatively is easier," he says, "but we need to define exactly how severe each threat is. We want to identify those issues that have the most practical significance so that we can prioritize them. For that, we need to be able to quantify the threats. We will also want to quantify the effectiveness of our countermeasures."
He anticipates that the most exciting aspect of the project will be evaluating the countermeasures and solutions. "We would like to do some partial implementation of our solutions. This is a novel problem that nobody has looked into yet." Finding the solutions now, he says, will make cognitive radio networks more robust and more reliable from the start.
For more information, visit www.arias.ece.vt.edu.