While potentially solving the spectrum underutilization problem using methods such as dynamic and opportunistic spectrum access, Cognitive Radios (CRs) also bring a set of security issues and potential breaches that have to be addressed. These issues come from the two important capabilities implemented within CRs: their cognition ability and reconfigurability. This chapter focuses on identifying, presenting, and classifying the main potential security attacks and vulnerabilities, as well as proposing appropriate counter-measures and solutions for them. These are supplemented by simulation results and metrics, with the intention of estimating the efficiency of each of the observed attacks and its counter-measure. The presented simulations are performed in the proprietary C/C++ and Matlab/Simulink simulators. nSHIELD is a major ongoing European embedded systems security-related project, which is used to demonstrate the practicability of the potential implementation of the proposed countermeasures and solutions for the discussed security problems and issues.
TopIntroduction
With the continuous market penetration of many spectrum-demanding radio-based services, such as video broadcasting, finding ways to increase the spectrum usage efficiency has become a necessity. Cognitive Radio (CR) is a technological breakthrough that – by utilizing concepts such as Opportunistic Spectrum Access (OSA) and Dynamic Spectrum Access (DSA) – is expected to be an enabler for these improvements, making it a current “hot topic” within the radio-communication research community.
Cognitive radio can be described as an intelligent and dynamically reconfigurable radio that can adaptively regulate its internal parameters as a response to the changes in the surrounding environment. Namely, its parameters can be reconfigured in order to accommodate the current needs of either the network operator, spectrum lessor, or the end-user.
Although this doesn’t necessarily need to be the case, Cognitive Radio (CR) is usually being defined as an upgraded and enhanced Software Defined Radio (SDR). Typically, full Cognitive Radios will have learning mechanisms based on some of the deployed machine learning techniques, and may potentially also be equipped with smart antennas, geolocation capabilities, biometrical identification, etc.
However, the newly-introduced cognitive capabilities are exactly what make Cognitive Radios susceptible to a whole new set of possible security issues and breaches. Furthermore, the threats characteristic to Software Defined Radios, as well as those characteristic to “traditional” wireless networks also need to be taken into account.
Cognitive Radio Network can be described as a network in which one or more users are Cognitive Radios. With the assumption of the potential attacker, as well as legitimate Secondary Users (SUs) always being CRs, the taxonomy of the threats within CRNs can be done with respect to the type of the Primary Users (PUs) considered, i.e.:
- •
PUs as “traditional” wireless systems,
- •
PUs as Software Defined Radios,
- •
PUs as Cognitive Radios.
It is worth noting that the proposed taxonomy is merely one of the possible approaches - the categorization can be done in several different ways. We have opted in for this particular approach because of its clarity and since it optimally fits the CR security framework that we are proposing in the last subsection.
The following subsection defines the basic concepts and premises of cognitive radios, and gives an introductory classification and basic definitions of CR-related threats. Section 3 will deal with the security of traditional wireless systems, describing legacy methods for protecting wireless communications such as WEP, WPA and WPA2, as well as the general security issues in wireless cellular networks. Section 4 highlights the security issues related to Software Defined Radio systems, while Section 5 considers potential threats to Cognitive Radios. Results are shown for so-called Primary User Emulation Attacks and Smart Jamming Attacks.
TopBackground
One of the most important capacities of future Cognitive Radio systems is their capability to optimally adapt their operating parameters based on the observations and previous experience. There are several possible approaches towards realizing such cognitive capabilities, such as:
Reinforcement learning refers to the machine learning method where radio learns through trial-and-error interactions in a scenario without perfect contextual information.
It is a kind of mathematical method used for the learning state in the cognition cycle, which will learn the information (recorded in the form of weighting factors) based on the external environment and previous states, which then influence the current activation. The weight is used to show the influence from the previous users or the factors based on circumstance, which will be updated on each activation (Hu, 2011).