Risk and Cost Multi-objective Optimization Using
Genetic Algorithm for Cognitive Radio Networks
By: Ismail Ahmed
Abstract:
The principle of Cognitive Radio Network (CRN) was introduced as a solution for the spectrum scarcity problem. Security of CRN holds a vital role in the availability of this technology in the future, however CRN is relevant to a high susceptibility to security threats due to its intelligent nature, learnability, and dynamic spectrum accessibility (DSA). Many security measures can be used nowadays, each of them has different implementation grades, and each grade has its own risk probability and implementation cost. It is important to achieve an optimum cost/risk efficient measures based on the available resources and the required security level. In this work a multiobjective security cost and security risk optimization problem is discussed and formulated to support
CRN decision in selecting the optimum security measures. A simplified case study is solved using both Brute Force method Modified Genetic Algorithm (GA). The results show that GA gives a very close answer to the optimal point.
Key words: Cognitive Radio Network, security cost, multi-objective optimization, genetic algorithm.
I.
Introduction
The spectrum scarcity problem has grown over the last decades due to the increase of wireless
devices usage and the spread of data-rich applications. According to the annual report of Cellular
Telecommunications Industry Association (CTIA) – The Wireless Association, in December 2013, the number of wireless devices such as laptops, notebooks, cellular phones, smart phones, and tablets in
USA exceeded the total number of people living in USA by 4.3%. Besides, the growth in data-rich consumer applications and wireless data transfer leaded to 219% increase in the annual wireless data
usage in 2013 in comparison to the wireless data usage in 2012 [1], that causes the USA’s frequency allocation chart to be very crowded, especially under 18 GHz.
Currently the spectrum is divided into two main categories: licensed and unlicensed users. In licensed spectrum, the designated users are allowed to utilize the resource exclusively, without fear of interference like, military, public safety, and commercial services like TV.
Unlicensed spectrum is licence-free frequency bands which are available for use by all users. Also, it is widely known that wireless device services degrade sharply when spectrum is crowded, for example, during emergency situations.
Cognitive radio (CR) technology was firstly introduced by Joseph Mitola III [2] as a possible solution for the spectrum scarcity problem. In CR technology, the users who are classified as Licensed users/Primary users (PUs), have the right to use the licenced spectrum band any time they want whereas unlicensed users/Secondary users (SUs) are defined as the users who can use the spectrum which is temporarily vacant, or not in use by primary users, without causing any violation the primary users communications capability.
Nowadays, Cognitive Radio Networks (CRN) applications range from smart grid, public safety, broadband cellular, and medical applications [3-5] to military applications especially in tactical mission scenario [6]. This wide range of applications leads to an increased susceptibility to security threats and attacks, such as jamming, eavesdropping, and denial of service attacks [7]. Security threats could target the short-term and long-term behaviours of the CRN, in short term behaviour CRN adapts incorrectly based on sensing of wrong environment information. The CRN learning ability and collaboration nature between nodes would propagate the error to the CRN long-term “future” behaviour. CRN intelligence nature in spectrum sensing, information sharing, and radio reconfigurability could be considered as double edge weapon because it could lead to vulnerability to new security
