Cognitive Engines:

Cognitive radio (CR) engines in literature are presented as the encephalon, the decision-making part of a cognitive radio system. They are often described as a multiple system of parameters that require delicate tuning to achieve optimum performance. There are numerous CR engines in the research arena that vary in many aspects depending primarily on the system that are to be employed while looking at network architecture, applications utilized, level of security required, hardware employed, radio environment, etc. Nevertheless, all variances of CR engines aim to achieve one thing, to optimise the performance of the system.

Performance in CR networks is defined in terms of multiple elements, such as bit error rate (BER), bandwidth, throughput, and transmit power. A significant number of CR engines have been inspired by Genetic Algorithms (GA) [1-4]. These algorithms are requiring modeling of the physical (PHY) layer traits of the radio within the context of a genetic chromosome. Biologically inspired CR engines are capable of intelligently adapting a radio’s physical and MAC behavior on constantly varying network conditions. An example were a GA driven CR decision engine is developed is presented in [5]. The GA based engine determines the optimal radio transmitter parameters for single and multicarrier systems. The GA in this case is employed to select optimal radio transmitter parameters for single carrier and multicarrier systems. The GA is based on a fitness function that directs evolution of the GA parameter to optimum. Further to the GA based CR engines, a
biologically inspired cognitive engine with dynamic spectrum access has been presented in [6]. Initial results indicate that the performance of the proposed engine achieved a 20dB SINR increase when compared with the traditional IEEE802.11 physical layer standard.  The wireless GA proposed in [6] is a multi-objective GA designed for the control of a radio by modeling the physical radio system as a biological organism and optimizing its performance through genetic and evolutionary processes. In [7], a prototype smart receiver has been presented including a General Purpose Processor (GPP) based software defined radio platform, signal classification capability and PHY-MAC re-configurability with hardware-independent radio interface. Another approach based on GA driven CR decision engines is outlined in [8]. This work explores the use of Taguchi method and orthogonal arrays (OA) as a tool for identifying favourable GA parameter settings. The strategies developed here limit the number of required tests needed to identify acceptable parameter values as opposed to the well-known methods such as design of experiment (DOE) and response surface methodology (RSM). As the number of configuration variables grows, DOE and RSM formalisations diminish due to the significant number of test cases that require gull factorial designs. Results presented in [8] indicate that the Taguschi method analysis yields a  redicted best combination of GA parameters from nine test cases. It utilises an efficient selection of testing configurations based on the concept of OA.

In [9] the theory and the prototypical implementation of the CE is presented. Furthermore a list of cognitive components is presented along with various issues related to developing algorithms for CR behaviour. These components are perception, conception and execution. According to [9] a cognitive engine interacts with the three main components of a CR architecture. The first one is the user domain which informs the cognitive engine performance requirements of services and applications to satisfy the minimum quality of service required. Based on the different QoS levels, the user domain effectively sets the performance goals of the radio. The second one is the external environment and RF channel deliver information regarding the changes in performance of waveforms based on different propagation environments. The third component is the generalized CR architecture is the policy domain that guides the system to perform within the
boundaries and limitations set by the local regulatory bodies as interpreted by the CE.

An important part of the design and development of CR engines are the tools employed to achieve the best possible result. In [10], a Policy Reasoner (PR) is proposed based on a language used for expressing policies that allow opportunistic spectrum access. Using a Policy Reasoner in CR engines, it is possible to guarantee policy-specified behaviours while allowing spectrum sharing. The proposed engine performs its processing at a software level than hardwired in the system like in legacy radios, thus achieving a device independent policy reasoner. The flexible mechanism coordinates, considers and processes large number of operating dimensions while adhere all regulation policies and maintain optimum spectrum sharing. The implementation of the Policy Reasoner in [10] was carried out using Cognitive (Policy) Radio Language (CORAL). CORAL has been first introduced in the neXt Generation (XG) programme funded by DARPA [11]. CORAL is a new language for policy specification that was devised to encode the policies in a straightforward,“natural” way in something close to first-order logic.The advantage of combining CORAL language with a device independent platform proposed in [10]was the capability to easily update the operating software based on various types of regulatory policies that exist. This would have an immediate effect on the operation of the engine in respect to:

1. Sensing frequencies (unrestricted bands, identification of primary users)
2. Characterisation of opportunities
3. Distributed or centralized coordination of resources based on communication with other devices for the identification of resource availability.
4. Enforcing behavior consistent with policies.

In [12] a cognitive radio architecture is presented while the authors discuss reasoning and learning engines as parts of
a CE. Reasoning and planning here are subjected to various radio parameters. With these parameters a knowledge-base is formed just like an expert system in Artificial Intelligence (AI) systems. The proposed engine is implemented based on a generic cognitive engine [13] which combines OSSIE [14] an open source software communications architecture SCA along with Soar Cognitive Engine.

[1] Zhijin Zhao; Zhen Peng; ShilianZheng; Junna Shang; , “Cognitive radio spectrum allocation using evolutionary algorithms,” Wireless Communications, IEEE Transactions on , vol.8, no.9, pp.4421-4425, September 2009
[2] Friend, D.H.; EINainay, M.Y.; Yongsheng Shi; MacKenzie, A.B.; , “Architecture and Performance of an Island Genetic Algorithm-Based Cognitive Network,” Consumer Communications and Networking Conference, 2008. CCNC 2008. 5th IEEE , vol., no., pp.993-997, 10-12 Jan. 2008
[3] Rondeau, Thomas W.; Le, Bin; Maldonado, David; Scaperoth, David; Bostian, Charles W.; , “Cognitive Radio Formulation and Implementation,” Cognitive Radio Oriented Wireless Networks and Communications, 2006. 1st International Conference on , vol., no., pp.1-10, 8-10 June 2006
[4] Si Chen; Wyglinski, A.M.; , “Cognitive radio-enabled distributed cross-layer optimization via genetic algorithms,” Cognitive Radio Oriented Wireless Networks and Communications, 2009. CROWNCOM ’09. 4th International Conference on , vol., no., pp.1-6, 22-24 June 2009
[5] T. R. Newman, B. A. Barker, A. M. Wyglinski, A. Agah, J. B. Evans and G. J. Minden, Cognitive Engine Implementation for Wireless Multicarrier Transceivers, Wiley Wireless Communications and Mobile Computing, 20 Sep 2006, http://www.ittc.ku.edu/publications/documents/newman2006_WileyJournal.pdf
[6] D. Maldonado, B. Le, A. Hugine, T. W. Rondeau, and C. W. Bostian, “Cognitive radio applications to dynamic spectrum allocation,” in IEEE International Symposium on New Frontiers in Dynamic Spectrum Access, 2005, pp. 597–600.
[7] B. Le, T. W. Rondeau, D. Maldonado, D. Scaperoth, and C. W. Bostian, “Signal recognition for cognitive radios,” in Software Defined Radio Forum Conf.Orlando, FL, 2006.
[8] Cognitive radio engine parametric optimization utilizing Taguchi analysis, A. E. Amanna, D. Ali, M. Gadhiok, M. Price and J. H. Reed, EURASIP Journal on Wireless Communications and Networking 2012, 2012:5
[9] T. W. Rondeau, “Application of artificialintelligence to wireless communications”,Ph.D. dissertation, Virginia Polytechnic Inst. and State Univ., Blacksburg, 2007
[10] Denker Grit, Elenius Daniel, SenanayakeRukman, Stehr Mark Oliver, and Wilkins David, “A Policy Engine for Spectrum Sharing,” Proceedings of the 2005 IEEE Symposium on New Frontiers in Dynamic Spectrum Access Networks (DySPAN 2007), Dublin, Ireland, 17-20 April 2007
[11] Grit Denker, Daniel Elenius, RukmanSenanayake, Mark-Oliver Stehr, Carolyn Talcott and David Wilkins, Cognitive Policy Radio Language(CoRaL)A Language for Spectrum PoliciesXG Policy LanguageVersion 0.1, ICS-16763-TR-07-00, April 2007, Prepared forDefense Advanced Research Projects Agency
[12] C. Clancy, J. Hecker and Erich Stuntebeck, Applications of Machine Learning to Cognitive Radio Networks, IEEE Wireless Communications, August 2007, pp.47-52
[13] E. Stuntebeck et al., “Architecture for an Open-SourceCognitive Radio,” SDR Forum Conf. 2006
[14] M. Robert, S. Sayed, C. Aguayo, R. Menon, K. Channak, C. V. Valk, C. Neely,T. Tsou, J. Mandeville, and J. H. Reed, OSSIE: Open Source SCA for researchers,”in SDR Forum Technical Conference, 2004.