SpecEES: Secure and Fair Spectrum Sharing for Heterogeneous Coexistent Systems

Summary and Project Goals

The dramatic growth in demand for wireless services has fueled a severe spectrum shortage, especially in the overcrowded unlicensed bands. The regulatory approach for meeting this galloping demand is to allow the coexistence of competing wireless technologies, cellular, Wi-Fi, radar, TV, emergency communications, and others. This shared spectrum paradigm poses novel challenges for the secure, efficient, and fair resource allocation. Many of these challenges stem from the heterogeneity of the coexisting systems, their scale, and the lack of explicit coordination mechanisms. Whereas some recent efforts have tried to address the coexistence of specific technologies, a comprehensive and general approach to securely and efficiently coordinate spectrum access for heterogeneous systems remains elusive. By facilitating the fair and secure coexistence of heterogeneous wireless systems, this project directly addresses the spectrum scarcity challenge. The outcomes of this project will benefit many industry verticals such as transportation, manufacturing, agriculture, critical national infrastructure, telecommunications, and others. Moreover, the expected results will advance knowledge in a number of scientific fields including security, privacy, and communication theory. Training opportunities for future wireless and security experts are also provided.

This project focuses on developing a novel coexistence framework for coordinating, monitoring, evaluating, and adapting spectrum access in a secure, efficient, and privacy-preserving manner. A general coexistence model in which spectrum can be horizontally and/or vertically shared in time, frequency, and space is considered. For this model, new fairness mechanisms that go beyond airtime sharing and account for unfairness due to spatial multiplexing are being investigated. These mechanisms dynamically allocate resources, not only in frequency and time but also spatially, yielding a fairer coexistence. Moreover, unfairness scenarios due to selfishly ignoring the coexistence etiquette are being modeled and assessed. The critical challenge is to detect and mitigate selfish misbehavior at the device and system level without explicit signaling. To overcome this challenge, the project explores implicit physical layer sensing and monitoring mechanisms that infer various operational attributes and safeguard spectrum access. Finally, the interplay between privacy, interference management, and efficiency in vertical spectrum sharing is investigated. The goal here is to protect the location privacy of assets deployed by legacy operators. Adaptive location obfuscation methods are being developed for achieving a desired degree of privacy while maximizing the spectrum efficiency.

Research Tasks and Progress

Thrust 1. LTE Misbehavior Detection in Wi-Fi/LTE Coexistence

The shared spectrum paradigm poses novel challenges for the secure, efficient, and fair resource access. Many of these challenges stem from the heterogeneity of the coexisting systems, the system scale, and the lack of explicit coordination mechanisms between them. The fundamentally different spectrum access mechanisms and PHY-layer capabilities--dynamic vs. fixed access, schedule-based vs. random access, interference-avoiding vs. interference-mitigating, etc.--create a complex and interdependent ecosystem, without a unified control plane. Figure 1 shows a typical co-existence model for LTE stations and Wi-Fi access points. Besides the standard differences, the two systems have different transmission powers that impact the respective carrier sensing mechanisms in different ways.

Figure 1  
Figure 1. Coexistence between LTE and Wi-Fi. Wi-Fi and LTE stations have difference interference ranges..
Some recent efforts have addressed the problem of fair coexistence of LTE/Wi-Fi and Wi-Fi/Zigbee under benign settings. However, the impact of deliberate violation of the coexistence etiquette to gain an unfair share of spectrum occupancy has not been studied at length. We were the first to study the impact of etiquette misbehavior under LAA-LTE and propose a mechanism for detecting such misbehavior.

Our methods build upon the extensive prior art on misbehavior detection in channel access for homogeneous networks, with notable differences. First, heterogeneous networks do not share common coordination channels for communicating explicit control information such as network allocation vector fields, device IDs, reservation messages (RTC/CTS), etc. Without explicit coordination, detecting the state and monitoring the behavior of stations operating under a different technology becomes challenging, as the messages exchanged by one system are undecodable at any other. Relevant challenges include determining which system occupies the channel, for how long, at what locality, with what range, which stations collided, to name a few. Moreover, although the LAA-LTE and Wi-Fi standards follow the same carrier-sense multiple access (CSMA) principles, they adopt different channel contention parameters that affect the overall system behavior under various conditions of coexistence. Determining a system's behavior requires accurate estimation of protocol parameters using only implicit monitoring. Note that Wi-Fi devices may not be equipped with LTE receivers and vice versa, thus complicating the monitoring mechanism.

We proposed a framework that enables the Wi-Fi to detect backoff misbehavior of LTE, taking into account the absence of any means for explicit coordination. Our framework relies on implicit sensing mechanisms that provide the Wi-Fi with accurate approximations of the backoff parameters used by the misbehaving LTE. Parameters of interest include the defer time before a channel access attempt, the backoff period for new and retransmitted frames, the LTE priority class, and the CW size. An overview of our framework is given in Figure 2.

Figure 2     Figure 3
Figure 2. Overview of the misbehavior detection framework. Figure 3. Detecting LTE transmissions using CP.
We were the first to study and formulate the problem of channel access misbehavior of LAA-LTE when coexisting with Wi-Fi. Although possible misbehaving strategies bear resemblance to those in a homogeneous setting, we highlighted novel challenges that stem from the technology heterogeneity and lack of explicit coordination

. We introduced a real-time monitoring mechanism that does not rely on message decoding for estimating relevant LAA-LTE protocol parameters. We develop an implicit sensing mechanism that goes beyond simple LTE transmission detection, to determining the existence of hidden stations, identifying retransmitted frames, and specifying the LTE priority class. These are essential parameters for accurately estimating the overall LTE behavior. A correlation -based mechanism for detecting transmissions of LTE stations without decoding is shown in Figure 3. Here, we exploit the time period of the cyclic prefix (CP) to pinpoint when an LTE station is active.

. We proposed a novel misbehavior detection mechanism based on Jensen-Shannon (JS) divergence. We analytically evaluated the threshold for detecting misbehavior based on the JS metric and characterize the detection and false alarm probabilities. We validated our theoretical results via extensive simulations and showed that our threshold selection mechanism yields near-perfect detection capabilities and negligible false alarm rates.

Figure 4 shows the ROC curve when the LTE station misbehaves half the time of the simulation by selecting a contention window between o and 3 (value of q = 4). We observe that the theoretical bounds are somewhat loose and that the true system performance is significantly better when the observation window (number of transmissions analyzed) is large. Indeed, the ROC is close to the optimal curve indicating that our system can operate with almost sure detection and almost zero false alarm probability. In Figure 5, we increased the value of the contention window q to 16 and repeated our simulations. Although the theoretical curve performs worse as the theoretical bounds become looser, the simulation results still demonstrate an almost perfect detection with an almost zero false alarm probability.

Figure 2     Figure 3
Figure 4. ROC curve when a = 0.5 and q = 4td> Figure 5. ROC curve when a = 0.5 and q = 16.

Thrust 2. On the Privacy and Utility Tradeoff in Database-Assisted Dynamic Spectrum Access

In dynamic spectrum access (DSA), commercially-operated database servers are often used to assist the opportunistic users (OUs) to query and access spectrum vacancies of incumbent users (IUs). The query and answer process in DSA architectures such as the one shown in Figure 6, introduce significant privacy concerns for potential leakage of the sensitive operational details of the IUs, especially their locations.

Figure 7  
Figure 6. DSA system architecture and query process.
Existing privacy-preserving mechanisms such as location cloaking or spatial obfuscation can be used to hide IUs’ locations. They, however, cannot guarantee the interference from all surrounding access-allowed OUs staying under a given limit. In addition, the spectrum utilization (i.e., transmission opportunities) of OUs should also be considered in the design of mechanisms. The complex three-way tradeoff among privacy, interference, and utility has not been systematically studied in the literature.

We are the first to define and quantify the three-way tradeoff among IU’s privacy, interference and OU’s utility for database-assisted DSA. We introduce the novel concept of placing a PZ inside of a EZ to protect the IU’s privacy and limit OUs’ interference. An example is shown in Figure 7. We define privacy as the adversary's probability of inferring IU's true location by observing the PZ and EZ, and OU's utility as the potential transmission opportunity (number of OUs that are allowed to transmit outside of the EZ). We consider two different scenarios. For deterministic OU location case (DOL), we formulate an optimization problem that maximizes the OU’s utility under a given interference constraint. We show that the problem is convex and present a simple algorithm to achieve the optimal transmission opportunity for OUs. For the case when OUs’ density is known, called the Probabilistic OU location case (POL) case, we formulate a non-linear optimization problem that optimizes either EZ’s or PZ’s radius under a non-linear interference constraint.

Figure 7  
Figure 7. (a) Privacy Zone (PZ); (b) Traditional Exclusion Zone (EZ); (c) PZ is contained in the EZ.
In addition, we have been developing new context-aware privacy notions that consider prior knowledge and randomized response based perturbation mechanisms which are applicable to location-data obfuscation. Also, we investigated location trace privacy-preserving mechanisms under the location-based services setting, where a user's locations are continuously updated and obfuscated and then released to an untrusted server. We leveraged information-theoretic notions to formulate the privacy leakage of the whole trace and define utility as the expected error of the released locations. Then we tried to derive the optimal utility-privacy tradeoff under certain utility constraint, by using tools from rate-distortion theory.

Team Members

Prof. Loukas Lazos (PI)

Prof. Marwan Krunz (Co-PI)

Prof. Ming Li (co-PI)

Mohammed Hirzallah (graduate student)

Bo Jiang (graduate student)

Ahmed Salama (graduate student)

Islam Samy (graduate student)

Broader Impacts

By facilitating the coexistence of heterogeneous systems in a fair, secure, and private manner, this project substantially improves the spectrum efficiency and security, and address the spectrum scarcity challenge. The outcomes of this project benefit many industry verticals such as transportation, manufacturing, agriculture, critical national infrastructure, telecommunications, and others. Moreover, the research agenda advances knowledge in a number of scientific fields including security, privacy, and communication theory. As an integral part of this project, the PIs provide training opportunities to graduates and undergraduates, as well as underrepresented minorities, to create the future security experts. Outcomes of this research are disseminated to a broader audience by integrating them into the curriculum, publishing in international venues, maintaining a project web page, and giving seminars nationally, as well as within our local community (through outreach activities).

List of Publications

  1. Zhang, Wenjing, Li, Ming, Tandon, Ravi, and Li, Hui, Online Location Trace Privacy: An Information Theoretic Approach pdf
    IEEE Transactions on Information Forensics and Security (TIFS), 14 (1) pp. 235 - 250, 2019.
     
  2. Jiang, Bo, Li, Ming, and Tandon, Ravi, Context-aware Data Aggregation with Localized Information Privacy pdf
    Proc. of the IEEE Conference on Communications and Network Security (CNS). pp. 1 - 9, 2018.
     
  3. Aronov, Boris, Efrat, Alon, Li, Ming, Gao, Jie, Mitchell, Joseph S., Polishchuk, Valentin, Wang, Boyang, Quan, Hanyu, and Ding, Jiaxin, Are Friends of My Friends Too Social?: Limitations of Location Privacy in a Socially-Connected World pdf
    Proc. of the Eighteenth ACM International Symposium on Mobile Ad Hoc Networking and Computing (MobiHoc), pp. 280 - 289, 2018.
     
  4. Salama, Ahmed, Li, Ming, Lazos, Loukas, Xiao, Yong, and Krunz, Marwan, On the Privacy and Utility Tradeoff in Database-Assisted Dynamic Spectrum Access pdf
    Proc. of the IEEE International Symposium on Dynamic Spectrum Access Networks (IEEE DySPAN). to appear. 2018.
     
  5. Hirzallah, Mohammed, Xiao, Yong, and Krunz, Marwan On Modeling and Optimizing LTE/Wi-Fi Coexistence with Prioritized Traffic Classes pdf
    Proc. of the IEEE International Symposium on Dynamic Spectrum Access Networks (IEEE DySPAN). to appear. 2018.
     
  6. Hirzallah, Mohammed, Afifi, Wessam, and Krunz, Marwan, Provisioning QoS in Wi-Fi Systems with Asymmetric Full-duplex Communications pdf
    IEEE Transactions on Cognitive Communications and Networking, 4(4). pp. 942 - 953, 2018.
     
  7. Ismaiel, Bushra, Abolhasan, Mehran, Ni, Wei, Smith, David, Fraklin, Daniel, Dutkiewicz, Eryk, Krunz, Marwan, and Jamalipour Abbas PCF-Based LTE Wi-Fi Aggregation for Coordinating and Offloading the Cellular Traffic to D2D Network pdf
    IEEE Transactions on Vehicular Technology, 67(12). pp. 12193 - 12203, 2018.
     
  8. Xiao, Yong, Hirzallah, Mohammed, and Krunz, Marwan Distributed Resource Allocation for Network Slicing Over Licensed and Unlicensed Bands pdf
    IEEE Journal on Selected Areas in Communications, 36(10). pp. 2260 - 2274, 2018.
     
  9. Samy, Islam, Lazos, Loukas, Li, Ming, Xiao, Yong, and Krunz, Marwan, LTE Misbehavior Detection in Wi-Fi/LTE Coexistence Under the LAA-LTE Standard pdf
    Proc. of the 11th ACM Conference on Security & Privacy in Wireless and Mobile Networks (ACM WiSec). pp. 87 - 98, 2018.