Cognitive radio: brain-empowered wireless communications

(Submited by jyothis T S, Lecturer in CSE, JEC Thrissur,

)

Introduction

Most of today radio systems are not aware of their radio spectrum environmentand operate in a specific frequency band using a specific spectrum access system.Investigations of spectrum utilization indicate that not all the spectrum is used inspace (geographic location) or time (see Figure 1). Therefore, a radio that can sense and understand its local radio spectrum environment is needed, toidentify temporarily vacant spectrum and use it, having the potential to provide higherbandwidth services, to increase spectrum efficiency and to minimize the need forcentralized spectrum management. This could be achieved by a radio that can makeautonomous (and rapid) decisions about how it accesses spectrum. Cognitive radioshave the potential to do this.

Figure 1: Spectrum usage

Consider a radio which autonomously detects and exploits empty spectrum to increase yourfile transfer rate. Suppose this same radio could remember the locations where your callstend to drop and arrange for your call to be serviced by a different carrier for those locations.These are some of the ideas motivating the development of cognitive radio. In effect, a cognitive radio is a software radio whose control processes leverage situational knowledge and intelligent processing to work towards achieving some goal related to the needs of the user,application, and/or network.

Arising from a logical evolution of the control processes of a software radio, cognitive radio presents the possibility of numerous revolutionary applications, foremost of which is opportunistic spectrum utilization. Cognitive Radio Technologies (CRT) was founded in 2007 by Dr. James Neel and Dr. Jeffrey Reed to speed the transition of cognitive radio from the laboratory to living room.With its extensive experience in the field of cognitive radio, CRT can help your products

  • Automatically detect and exploit unused spectrum
  • Automatically detect and interoperate with varying network standards
  • Improve performance.

There are many definitions of CR and definitions are still being developed both in academiaand through standards bodies, such as IEEE-1900 and the Software Defined Radio Forum.Summarizing Mitola, a full CR can be defined as “…a radio that is aware of its surroundingsand adapts intelligently”. This may require adaptation and intelligence at all the 7 layers ofthe ISO model.

A working definition used is:

“A CR uses intelligent signal processing (ISP) at the physical layer of a wireless systemand is achieved by combining ISP with software defined radio (SDR)”.

In this working definition a CR makes use of a flexible radio and intelligence so that it canadapt to changes in the environment, to its user’s requirements and to the requirements ofother radio users sharing the spectrum environment. Cognitive radio is viewed as a novel approach for improving the utilization of a precious natural resource: the radio electromagnetic spectrum. The cognitive radio, built on a software-defined radio, is defined as an intelligent wireless communication system that is aware of its environment and uses the methodology of understanding-by-building to learn from the environment and adapt to statistical variations in the input stimuli, with two primary objectives in mind:

  • Highly reliable communication whenever and wherever needed
  • Efficient utilization of the radio spectrum.

For these, a complete CR node solution with an intelligent layer ofawareness, reasoning and learning necessary to optimizeperformance under dynamic and unpredictable situations is needed.Such an intelligent layer is realized by a software systemcalled a cognitive engine (CE). The CE can be applied todifferent reconfigurable radio platforms via its general radiointerface. The CE embeds a two-loop cognition cycle as itslearning core. The cognition cycle integrates radioenvironment sensing and recognition, case-based reasoningand solution making, and evolutionary solution improving.Radio knowledge is defined and the knowledge database isimplemented to support the reinforcement learning throughthe cognition cycle.To be generally applicable for various applications, theCR solution emphasizes platform independent systemarchitecture, and the CE has an algorithm framework that isopen-structure and modular, which can be easilyreconfigured for the target problem.Based on this general CR node structure, a fullyfunctionalpublic safety cognitive radio (PSCR) node isprototyped to provide the universal interoperability forpublic safety communications. The complete PSCR nodesoftware system has been packaged for outsideorganizations to build prototypes and carry on field testing.

AFull CR is assumed to be a fully re-configurable radio device that can cognitivelyadapt itself to both user’sneeds and its local environment. Forexample, a mobile handset may use cognitive reasoning to automaticallyreconfigure itself from a cellular radio to a PMR radio, or it may automatically powerdown when in a sensitive environment (such as a hospital, cinema or airport). Thisfull CR is often referred to as a Mitola radio (named after the MITRE scientistJoseph Mitola). It is unlikely to be achieved in the next 20 years because it impliesthe availability of full software defined radio technologies coupled with cognitivecapabilities. If flexibility of hardware and intelligence to control or configure thehardware, are two axes of a matrix (see Figure 2), then a full cognitive radio(Mitola radio) would be at the top right.

Figure 2: A matrix with full cognitive radio

What is Cognitive Radio?

According to Mitola, who first coined the term, a cognitive radio should findavailable bandwidths and filter out unnecessary information. It will be clever aboutwhat the user wants and will know how to get the right information to the user in anefficient manner. It will do this automatically without bothering the user. The four most popular emerging interpretations of CR are:

  • Full Cognitive Radio - also called Mitola Radio, in which every possibleparameter observed by the radio is taken into account while making adecision on the way it operates.
  • Spectrum Sensing Cognitive Radio - in which only radio frequency(RF) spectrum is observed and consequently used in decision making.
  • Licensed Band Cognitive Radio - in which the device is capable ofusing licensed spectrum in addition to unlicensed spectrum.
  • Unlicensed Band Cognitive Radio - in which the device is allowed touse license exempt and/or free license spectrum only.

Thus a cognitive radio as one that usesintelligent signal processing at the physical layer of a wireless system.CR is the amalgamation of software defined radio (SDR) and intelligent signalprocessing (ISP). Combining the facets of radio flexibility, intelligence and spectral awareness, a full CR will adapt itself to changes in the environment, its user requirements and the requirements of other radio users sharing the spectrum (in time and space). A full CR will also use long-term analysis to learn about its environment and its own behavior. CR implies intelligent signal processing (ISP) at the physical layer of a wireless system, i.e. the layer that performs functions such as communications resource management, access to the communications medium, etc. Usually, (but notnecessarily) it is accompanied by ISP at higher layers of the Open System Interconnection (OSI) model. If ISP is not implemented at thesehigher layers then a CR will be restricted in what it can do. Because acommunication exchange uses all seven OSI layers, ideally all seven layers need tobe flexible if the CR intelligence is to be fully exploited. Without optimization of allthe layers, spectrum efficiency gains may not be optimized. This level of complexity,required for the full (Mitola) CR, may not be achievable for many years.

The Dimensions of a Cognitive Radio:

The two key technologies required to make a CR provide the twoessential characteristics that make a radio cognitive. These are flexibility (providedby SDR) and intelligence (provided by ISP). These two factors may be exhibited atvarious levels of complexity and/or ability. This is why CR is hard to define: instead,there will be generic capabilities of CR ranging from the most basic adaptation to themost advanced (e.g. a Mitola radio).

A matrix based on RF flexibility and intelligence can help clarify the varying gradesof CR, see Figure 3. Anadvanced form of CR cannot exist without both factors. A device may have the veryhighest level of intelligence but without the RF flexibility to tell it about theenvironment (for example a wideband antenna), it cannot make informed decisions.Conversely, an extremely flexible device is not worth much if it lacks the intelligenceto make use of the information it is receiving.

Figure 3: Matrix concept for grading CR

Thus Figure 3 shows that RF flexibility and intelligence must both increase to attainan advanced form of CR.

Main functions:

The main functions of Cognitive Radios are:

  • Spectrum Sensing - detecting the unused spectrum and sharing it without harmful interference with other users. It is an important requirement of the Cognitive Radio network to sense spectrum holes. Detecting primary users is the most efficient way to detect spectrum holes. Spectrum sensing techniques can be classified into three categories:
  • Transmitter detection: cognitive radios must have the capability to determine if a signal from a primary transmitter is locally present in a certain spectrum, there are several approaches proposed:
  • Matched filterdetection
  • Energy detection
  • Cyclostationaryfeature detection
  • Cooperative detection: refers to spectrum sensing methods where information from multiple Cognitive radio users are incorporated for primary user detection.
  • Interference based detection.
  • Spectrum Management - Capturing the best available spectrum to meet user communication requirements. Cognitive radios should decide on the best spectrum band to meet the Quality of service requirements over all available spectrum bands, therefore spectrum management functions are required for Cognitive radios, these management functions can be classified as:
  • spectrum analysis
  • spectrum decision
  • Spectrum Mobility - is defined as the process when a cognitive radio user exchanges its frequency of operation. Cognitive radio networks target to use the spectrum in a dynamic manner by allowing the radio terminals to operate in the best available frequency band, maintaining seamless communication requirements during the transition to better spectrum.
  • Spectrum Sharing - providing the fair spectrum scheduling method. One of the major challenges in open spectrum usage is the spectrum sharing. It can be regarded to be similar to generic media access control MAC problems in existing systems.

Cognitive radio architecture:

An algorithm software package, called the cognitiveengine (CE), is designed and overlaid on the radio hardwareplatform. The CE manages radio resources to accomplishcognitive functionalities and adapts radio operation tooptimize performance. The CE enables aradio to provide cognitive functionalities by combining themachine learning process with radio operation.


Figure 4: Cognitive radio system model

A machine learning core is designed to enable cognitivecapabilities for wireless applications. Reinforced learningand evolutionary optimization are key design principles ofthe learning core. A two-loop cognition cycle is embedded in the learning core. Any radio with an appropriate level of reconfigurabilitycan support and be controlled by the CE via a platformindependent radio interface. Since CE is not platformspecific, general knowledge and learning can be applied fora variety of applications’ problems.The cognitive functionality focuses on layers 1 to 3 toachieve cross-layer optimization. The general cognitionalgorithms can be extended to higher layers, and configuredto meet various application specific requirements.As a network node by nature, a CR can work individuallyor jointly on resource management and performanceoptimization. The CR learning structure consists of threesteps: recognition, reasoning and adaptation, which can beflexibly implemented in either a centralized way as a fullyfunctional CR node or be distributed across the networkwhere different local parts of the network require differentlevels of intelligence and different layers of optimization.Such CR node functional structure is shown in Figure 5.

Figure 5: CR functional structure as a network node

Cognitive radio (CR) versus intelligent antenna (IA):

Intelligent antenna (or smart antenna) is antenna technology that uses spatial beamforming and spatial coding to cancel interference; however, it requires intelligent multiple or cooperative antenna array. On the other hand, cognitive radio (CR) allows user terminals to sense whether a portion of the spectrum is being used or not, so as to share the spectrum among neighbor users. The following table compares the different points between two advanced approaches for the future wireless systems: Cognitive radio (CR) vs. Intelligent antenna (IA).

Point / Cognitive radio (CR) / Intelligent antenna (IA)
Principal goal / Open Spectrum Sharing / Ambient Spatial Reuse
Interference processing / Avoidance by spectrum sensing / Cancellation by spatial pre/post-coding
Key cost / Spectrum sensing and multi-band RF / Multiple or cooperative antenna arrays
Challenging algorithm / Spectrum management tech / Intelligent spatial beamforming/coding tech
Applied techniques / Cognitive Software Radio / Generalized Dirty-Paper and Wyner-Ziv coding
Basement approach / Orthogonal modulation / Celluar based smaller cell
Competitive technology / Ultra wideband for the higher band utilization / Multi-sectoring (3, 6, 9, so on) for higher spatial reuse
Summary / Cognitive spectrum sharing technology / Intelligent spectrum reuse technology

Table 1: Cognitive radio (CR) vs. Intelligent antenna (IA).

When will CR happen?

Full Cognitive Radios (Mitola radios) do not exist at the moment and are not likely to emergeuntil 2030, when fully flexible SDR technologies and the intelligence required to exploit themcognitively can be practically implemented.However, true cognition and fully flexible radios in terms of the Mitola definition may not beneeded, as simple intelligence and basic reconfigurability at the physical layer could providesignificant benefits over traditional types of radio.

There are two main obstacles to realizing a Full CR. The first is the challenge ofmaking a truly cognitive device, or a machine with the ability to intelligently makedecisions based on its own situational awareness. The second challenge is reliance onthe development of SDR technologies to enable reconfigurability. It is expected thata single full CR (Mitola) device capable of operating in any frequency band up to3GHz without the need for rigid front-end hardware (excluding the antenna) will notbe available before 2030.Within the next five years CR prototypes will have emerged and perhapseven one or two market products will be available. These will rely heavily ondevelopments in SDR. They will not be very intelligent and will use logical andanalytical ISP rather than cognition.

Techniques within reach of Current Technology

While no-one has built a fully-fledged cognitive radio, there are manycommunications devices in use today that exhibit some of the characteristics of aCR. For example adaptive control of transmit power, spectrum allocation, networkaccess and spatial allocation can be found, to varying degree, in a number ofexisting devices. Cognitive stacks are well-established with capabilities includingautonomous variation in modulation schemes, coding, network routing and radioresource management. Some examples include:

  • Adaptive Power Control
  • Adaptive Spectrum Allocation
  • Adaptive Modulation
  • Adaptive Coding
  • Adaptive Network Access
  • Adaptive Routing
  • Adaptive Spatial Allocation
  • WCDMA Resource Management
  • Adapt4 Cognitive Radio

The degree to which current technology is capable of cognitive radio behaviour canbe mapped onto the matrix of RF flexibility and ISP, as shown in Figure 6. As thereare no full software radios in existence the top two rows remain empty. Similarlythere is no machine capable of intuition, so the right-hand column also remainsempty. This leaves one-third of the matrix to be populated with today technology,most of which sits in the bottom left division. As the matrix shows, there is still a longway to go before a Mitola radio is achieved.

Figure 6: Matrix of CR technology available today

Techniques Expected in the Future

In the near future, that is in a next few years, the followingdevelopments are expected:

  • Cognitive Analysis
  • Software Defined Radio
  • Impulse Radio (UWB)

Looking further into the future, beyond 25 years time, the following developmentsare expected:

  • Machine Intuition
  • Full Software Radio

What are the potential applications of CR andwhat spectrum could it use?

CR techniques whichallow spectrum sharing with other spectrum users are ideal for non-time critical applications.Four promisingapplications identified are:

  • Mobile multimedia downloads (for example, download of music/video files to

portable players) which require moderate data rates and near-ubiquitouscoverage;

  • Emergency communications services that require a moderate data rate and

localised coverage (for example, video transmission from firemen’s helmets);

  • Broadband wireless networking (for example, using nomadic laptops), which

needs high data rates, but where users may be satisfied with localised lot spot services;

  • Multimedia wireless networking services (e.g. audio/video distribution within

homes) requiring high data rates.

A number of applications were identified that could exploit CR and anumber of bands where CR could co-exist were highlighted. Detailed research isessential to test the potential impact of sharing and how capable two networks reallyare of co-existing in the same spectrum band. Additional applications are constantlyemerging as CR technologies develop.

The advantages and disadvantages of each are summarized in Table 2.

Table 2: Advantages & disadvantages of most suited applications for sharing with

CR

What are the key benefits and challenges ofCR?

The main specific benefit of full CR is that it would allow systems to use their spectrumsensing capabilities to optimise their access to and use of the spectrum. From a regulator’sperspective, dynamic spectrum access techniques using CR could minimise the burden ofspectrum management whilst maximising spectrum efficiency.Additional benefits from the development of SDR, coupled with basic intelligence, are:optimal diversification enabling better quality of service for users and reduced cost for radiomanufacturers.There are three main challenges to the widespread deployment of CR. First, ensuring thatCRs do not interfere with other primary radio users – i.e. solving the hidden node problem.Second, because CR relies on SDR, all the security issues associated with SDR, such asauthenticity, air-interface cryptography and software certification etc, also apply. The thirdchallenge is control of CRs. It is not clear how, or if, these problems can be solved.

The benefits of CR were identified as:

  • optimal diversity
  • spectrum efficiency
  • commercial exploitation
  • quality of service

How will CRs be controlled in a changing radiospectrum environment?

CRs by their nature will be very flexible and have the potential to interfere with other users of shared radio spectrum. Their behaviour, therefore, must be controlled oragreed in some way. Because the greatest cause for concern lies with how tochoose the correct carrier frequency, the consortium focussed on briefly examiningpotential spectrum control methods. A number of such methods exist andoperationally they may be band-specific. For the purposes of this study a PMRscenario with increasing complexity was developed, to explore the impact of differingspectrum control techniques. Three main techniques were considered: