Wireless Regional Area Networks and Cognitive Radio(1)

Abstract:IEEE has set up in October 2004 the IEEE802.22 Working Group?Wireless Regional Area Network (WRAN) Task Force to work out air interface standard based on Cognitive Radio (CR) technologies. The standard includes Physical Layer (PHY) and Media Access Control (MAC), to use the already allocated fallow spectrums to broadcast TV in a non-interference way. The WRAN employs CR technologies to sense and estimate the television frequencies and use the technologies of dynamic spectrum management to find and then allocate idle spectrums. The CR technologies are representing one of the major trends for future wireless communications. This article on WRANs and CR technologies will be divided into two issues. In this issue, WRANs and IEEE 802.22, CR technologies are introduced. And the second part in the next issue will analyze the applications of CR technologies.

1 Wireless Regional Area Networks (WRANs) and IEEE 802.22
The advent of information era has been followed by all of the more extensive application of wireless services and devices, such as mobile communications, radio broadcast, and television. All these are exhausting spectrum resource and making it more precious than land and mineral resources. Nowadays, transmitters and receivers for communication systems need certification from government to enter a communication network. Therefore, there are breakthroughs in the unlicensed frequency bands such as 2.4 GHz?2.484 GHz for Wireless Local Area Network (WLAN). Likewise, cutting-edge technologies come forward and are driving the standard creating organizations to ponder over the current spectrum usage and possible methods that boost utilization efficiency of radio resources.
In November 2002, the Federal Communications Commission (FCC) issued a report submitted by Spectrum Policy Task Force (SPTF) on improving spectrum management. For many frequency bands, the spectrum access is more significant problem than physical scarcity of spectrum. Spectrum scanning could easily reveal an embarrassment that lies in our urban and remote rural areas: some frequencies are left idle for most of the times while others are overused. Knowing this, FCC has started reexamining traditional spectrum management methodologies.

      Cognitive Radio (CR) technologies have come and been accepted as the possibly optimum solution to improve spectrum efficiency. The CR system is an intelligent wireless communication system that is aware of its surrounding environment. It learns from the environment and adapts its internal states in real time to statistical variations of radio parameters, such as power, carrier modulation, and coding. These technologies can lead to the advent of smarter unlicensed devices that make greater use of spectrum than possible today?without interfering with licensed users or Primary Users (PUs) for highly reliable communications whenever and wherever. The CR is now being developed fast with the help of other related technologies: signal processing, artificial intelligence, Software Defined Radio (SDR), frequency agile, and power control.

      In May of 2004, FCC issued a Notice of Propose Rulemaking (NPRM) that allows an unserviced user to "borrow", through CR technologies, a licensed yet fallow spectrum originally allocated for TV broadcast purposes. He has to make sure his traffic would not interfere with the normal traffic of the serviced user (TV viewer) of this spectrum.

      In October of 2004, the IEEE802 Committee had set up the IEEE802.22 Working Group. It磗 developing functional requirements for a Point-to-multipoint (P-MP) interoperable air interface of PHY and MAC. The standard applies to the current spectrum of TV broadcast services to implement CR-based Wireless Regional Area Network (WRAN).

      The following paragraphs will give a fully detailed view of IEEE 802.22, including its target market, systems, protocols, and air interface applications.

1.1 Target Market
The IEEE 802.22 protocol-based WRAN is now applied largely in rural, remote and scarcely populated areas. It makes use of the unused TV broadcast frequencies for wireless broadband access that could be left difficult otherwise, on condition that no interference is incurred thereby. However, it doesn磘 necessarily mean that the IEEE 802.22 has its applications restricted to remote countryside, it has other target market objectives as well, including single family, residential, multi-dwelling units, SOHO, small businesses, multi-tenant buildings, public and private campuses. The definition of IEEE 802.22 system also makes it applicable in urban and suburban areas. It磗 believed that CR-based technologies will bring new excitements to today磗 wireless communication systems.

1.2 IEEE 802.22 System
An IEEE 802.22 system is implemented on its PHY and MAC layers, and at least one user communicates with the base station via the P-MP wireless air interface. The P-MP radio aims to use the TV VHF/UHF frequencies.

      Figure 1 below shows the schematic diagram of an IEEE 802.22 WRAN system. The system should have at least one base station and one Customer Premier Equipment (CPE). The base station works in P-MP mode. It uses an omni antenna and a formed sector or adaptive antenna array to transmit downlink signals to CPE. For the sake of coexistence with primary user, the system should include proper PHY and MAC mechanisms for the base station to sense the primary user磗 frequency and dynamically change the network power or frequency to avoid interference. For better spectrum sharing, the system should include a mechanism to coordinate relations between base stations when there are coverage problems, for example, overlapped coverage.

1.2.1 Reference Model for Wireless Access
Figure 2 shows the radio access reference model of the IEEE 802.22 system. It describes the air interface model between User Networks (UN) and Core Networks (CN). The large-scale of IEEE 802.22 system should normally comprise user terminal, base station interconnection network and network management devices. However, the model given here focuses on its air interface aspect, for example the Core Network Interface (CNI) and User Network Interface (UNI). A CPE is able to support several users via UNI and the users can transmit data, voice and image in between. Likewise, a base station is able to support several CNs via one or more CNIs.

1.2.2 Network Entity Relation and Topology
The IEEE 802.22 system defines invariable P-MP air interfaces (as shown in Figure 1). The repeater is used to expand coverage or network capacity. All devices are put under the control of base station, including centralized power control, spectrum management and scheduling.

      The base station monitors and controls the radio resource allocation. The base station controls the media access and transmits downlink information to CPEs, while the CPEs respond to the base station in uplink direction. To protect licensed services, the IEEE 802.22 system defines strict master/slave relation between the base station and CPEs. Therefore, no CPE is allowed to transmit information until it磗 authorized to do so by the base station. The base station also controls the Radio Frequency (RF) features (modulation, coding, and working frequency) of all CPEs. Besides, a base station in an IEEE 802.22 system features distributed sensing management to double protect licensed services. By distributed sensing it means that the base station controls CPEs remotely to make them sense the RF environment. Once there is a feedback, it uses the feedback information to decide what next steps should be taken by the sensing unit.

1.2.3 Service Capacity and Coverage
In a P-MP IEEE 802.22 system, the base station should have big enough capacity to provide services to numerous CPEs. Each CPE needs a minimum downlink throughput of 1.5 Mb/s and an uplink peak throughput of 384 kb/s. Spectrum efficiency is an important performance index, as 0.5 bit/(sec/Hz)-5  bit/(sec/Hz) in IEEE 802.22 system. If reasonably designed, each CPE on a possible transmission link can communicate with maximum spectrum efficiency.
Compared with the ready-made IEEE 802 standards, the IEEE 802.22 WRAN features wider coverage (up to
40 km-100 km) of base stations. Figure 3 shows the coverage of different networks.

1.3 IEEE 802.22 Protocol
The IEEE 802.22 protocol defines coordinated operation of multiple devices. Protocol interaction exists in every single layer of the protocol. The IEEE 802.22 MAC and PHY protocols are the same for all supported services. The major objective of MAC layer is to share wireless channel resources. The MAC protocol defines how and when to initialize transmission on the channel.

      As CPEs have to compete for capacity of one or more base stations, the MAC protocol should manage the competition and allocate resource effectively. The PHY layer can be sub-divided into a convergence sublayer and a Physical Medium Dependent (PMD) sublayer. The PMD sublayer takes the major part of the PHY layer. As the convergence layer of MAC, the convergence layer of PHY is able to adaptively map specific MAC requirements to universal PMD services.

1.4 IEEE 802.22 Air Interface
Because it磗 the TV frequency bands that are used by P-MP wireless devices in an IEEE802.22 system, base station and user may communicate with each other in a short-range environment and Non-line of Sight (NLOS) environment. For co-existence with primary user, the PHY and MAC layer protocols of IEEE 802.22 should allow the base station to adjust dynamically the system磗 power and working frequency based on what it senses. A noise control mechanism is also necessary to prevent interfering with primary users of TV frequencies.

      To work without affecting licensed services, the IEEE 802.22 air interface requires relatively strong adaptability and expandability. The adaptability means firmware and software downloading to modify specific transmission parameters and update CPE parameters. It includes the adaptability of rate and power. The expandability refers to the changeability of system磗 operating parameters. It includes the expandability of bandwidth and link symmetry.
Following, we磍l look at the PHY and MAC architectures that support the above two features. This is also necessary for coexistence with primary user.

1.4.1 The PHY Layer
The randomness when there is idle frequency for the IEEE 802.22 base station to communicate with CPE has a bearing on the architecture of PHY and MAC layers. The PHY should be designed to allow high-performance reliable communication at the cost of relatively low complexity.

      The IEEE 802.22 PHY layer features highly expandable modulation and encoding. It enables a base station to dynamically adjust modulation and encoding mode to agree with varying Signal-to-noise Ratios (SNRs) that are caused by different distances from CPEs to the base station. System efficiency can be boosted hence.

      The IEEE 802.22 standard supports efficient Transmit Power Control (TPC) of links. It allows a CPE to reduce its power to the minimum level that is just enough for maintaining a link reliably. To maximize link throughput, a compromise can be made between relatively low transmit power and flexible modulation mechanism. For coexistence磗 sake, PHY should consider Dynamic Frequency Selection (DFS), to adjust operating frequency as fast as possible and energy-efficiently.

1.4.2 The MAC Layer
To respond promptly to environment changes, for example, appearance of a primary user, the CR-based IEEE 802.22 MAC layer should feature highly dynamic functions. Besides, traditional services such as media access control and robust data transmission, it is supposed to provide a suite of brand new functions. That is including distributed spectrum sensing and dynamic spectrum management, in a bid to share spectrums with TV and radio.

      (1) Access Initialization
      As defined by IEEE 802.22 MAC protocol, a started CPE first spends some time on scanning all TV frequencies to set up an identifier. That is a mapping diagram of each channel磗 utilization status, which shows whether the primary user signal is detected or not. The result will then be sent to the base station. This differs from current radio technology where a CPE uses pre-decision channel to search for base station. Besides, IEEE 802.22 may also use channel convergence technology to combine multiple idle channels for higher performance.

      (2) Sense Management and Spectrum Management
      To prevent an IEEE 802.22 system from devastatingly interfering with primary users, the base station should direct CPEs to periodically sense and estimate the in-band or out-band environment. Out-band estimation is targeted to all unaffected channels while the in-band estimation evolves channels the base station uses to communicate with CPEs and adjacent channels that may be affected by such communication. While in-band estimation is being performed, the base station has to stop data transmission inside the channel, which is unnecessary for out-band estimation however.  An IEEE 802.22 device recognizes a primary user through signal sensing in a non-coherent way and under base station磗 dynamic control when the SNR is relatively low. Howewer, since the base station cannot communicate with CPEs while in-band sensing is in progress, longer sensing period will end up in greater degradation of the communication performance.

      In addition, not all CPEs have to carry out the same sensing task. Instead, the base station uses some intelligent composite algorithms to allocate sensing tasks to various CPEs. Once it obtains enough sensing data, it integrates the data to form a spectrum status diagram of a cellular unit. It then takes proper steps to change related CPE operating parameters such as operating frequency and power level to prevent possible interference.  The MAC layer design should also include the duration and frequency of sensing estimation, as well as the device to be adopted.  An IEEE 802.22 system also contains a package of spectrum management functions: switch channel, suspend/restart channel transmission, and terminate/restart channel operation. All these functions aim to protect primary users and ensure coexistence of primary users with unserviced users.

      (3) Propagation Delay
      Another important issue to be addressed by MAC design is the support of propagation delay. When the IEEE 802.22 system attempts to serve in a 100 km range around the base station, the circular propagation delay exceeds 30 ms. Such limit the MAC layer has to compensate for varied propagation delays caused by different CPEs. Too long propagation delay would tamper with effective access, which is very important for the co-existence of various services.

2 CR?Makeing Radios Smart
The CR plays an essential role in WRAN. It helps reuse the existing TV spectrums and improves current use efficiency of radio spectrum by means of sensing/estimating TV frequencies, and managing spectrums dynamically. Although still in its toddling stage, CR is believed to have the potential to enhance radio spectrum access efficiency and boost radio communication performances.
In the following sections we磍l look into the research work, basic functions and SDR-based implementation of CR, and Radio Knowledge Representation Language (RKRL), sensing circulation, CR磗 etiquette, and radio sensing network.

2.1 Status Quo of CR Research
The term "cognitive radio" was coined by Joseph Mitola in his doctoral dissertation in 1999. Mitola described how a cognitive radio could enhance the flexibility of personal wireless services through a new language called the RKRL.
In November 2002, FCC issued the SPTF report and set the date (May of 2003) to establish the CR work group in Washington. In March 2004, a CR symposium was held in Las Vegas that signified the first step in CR technology development.

      The academic world didn磘 just wait and see either. A well-known communication theory specialist Simon Hakin had a CR overview article published in the February issue of JSAC in Communications in 2005: Cognitive Radio: Brain-empowered Wireless Communications, which had led the global academic society into CR study.

      Subsequently, graduate schools in Berkeley, Virginia and Stevens universities and research institutions/forums such as SDR Forum have joint the legion of CR research work. The Winlab of Rutgers University is currently working on the CR platform.Presently, the Defense Advanced Research Projects Agency (DARPA) of the US Department of Defense is engaged in a Next-generation (XG) communications program. The program plans to work out CR-centered system methodology and key technologies to enable dynamic spectrum access and sharing. The DARPA claims to improve current spectrum efficiency by 10-20 times as much. Intel, Qualcomm, Philips and Nokia and some other big names have also embarked on CR research.

      To use CR for better spectrum efficiency not only takes technological approaches, it also requires regulation support from governments. In May 2004, the US promulgated Notice of Proposed Rule Making (NPRM). It allowed the use of CR-based technologies to access unlicensed radio resources of broadcast TV airwaves on condition of not restricting the privileges of primary users (for example, the TV viewers). Japan has also declared that it would start CR research in 2005. In Europe, Germany, Britain, Italy and Sweden also have research organizations working on CR.

      Besides, the international society has seen two important seminars on CR-related technology and dynamic channel allocation held in the past two years. One is Cognitive Radios Conference held in October 2004 in Washington, D.C. The other is Dynamic Spectrum Access (DySPAN) held in November 2005, on dynamic spectrum allocation and access technologies. The DySPAN conference published over 80 papers.

2.2 Basic Functions of CR
The key words we can get from CR磗 definition are sense, intelligent, learning, adaptive, reliability and efficiency. The sensing feature of CR covers waveform sensing, spectrum sensing, network sensing, geo-location sensing, locally available service sensing, user requirements sensing, language sensing, status sensing as well as security policy sensing. Spectrum sensing has drawn the most attention so far. Intelligence is the most important feature that makes CR extraordinary if compared to other SDRs. The CR is an objective-driven framework and it responds to events in a sequence of planning, decision-making and execution.

      The CR is able to observe the radio environment, analyze information, evaluate selections, produce plans and monitor services. It obtains specific capabilities via paradigm-based reasoning. The CR is able to learn from mistakes, including mechanical learning with or without instructions. Such a circulation of observation, reasoning and action is totally different from today磗 mobile phones, as the latter exchanges data and execute network commands only on frequencies specified by user.

      To sum up, CR would need programmable software radio platform, as well as robust computer model, user model, network model and RF transmission environment.

2.3 RKRL and SDR-based CR
The CR expands SDR functions through modeling in the wireless field. It brings more flexibility to personal services with the help of RKRL. The language representation consists of wireless mode, device, software module, transmission, network, and user needs, as well as application mode that is automatically configured to meet user requirements. The RKRL can work on the SDR platform, which then turns wireless nodes from executing pre-defined protocols into intelligent agent of wireless domains: a reform of systems from carrying invariable function and communication model into intelligent communication systems.

      With RKRL, a CR system knows the parameters for advanced language to configure equalizer interface and interface line delay structure. Through its internal architecture model, RKRL can address problems in proper modes. That is, the network is able to raise random problems in standard RKRL, while any RKRL can reply to the problem. Normally, CR doesn磘 need human intervention in its work, as it磗 a complex adaptive system therefore can adapt to outside environment changes.

2.4 RKRL
As far as CR is concerned, the way of representing outside information is very important. The RKRL is a suite of language architecture developed at the Royal Institute of Technology, Sweden, to describe knowledge, plan and requirements of the whole world. It includes the Knowledge Query and Manipulation Language (KQML) to make internal knowledge exchange more convenient.

      The RKRL provides a standard language that is able to dynamically define random data exchange. Every part of the RKRL is composed of rule mode and descriptive language basis, with their capability derived from the configured mode, sensing circulation speed and related interfaces？ intervention speed.  The RKRL includes syntax and text information and it covers rule mode, language definition, deduction mode, multi-syntax and the radio itself.  The RKRL is a parallel-object language and it combines features of above languages through mode-based reasoning.
With the standard language of RKRL, it is possible to dynamically define burst data conversion in a CR system and the agent is able to make radio rules better serve users through fast protocol operations. With these operations enhancing the system flexibility and reaction capability.

2.5 CR Rules
The CR works just as a game as it proactively and freely selects RF channel, air interface, protocol architecture, and service price (to compete with other users). The network may coordinate such game rules. Many researchers would adopt the game theory for channel spectrum allocation and power control. For balanced use of radio resource in the future, a CR system will be fully capable of competing with other systems in some frequency bands.

2.6 Sensing Circulation
As an intelligent wireless communication system built on SDR platform, CR uses the methodology of
understanding-by-building to learn from the environment and adapts its internal states (including transmit power, carrier frequency, and modulation mode) accordingly.  It does it with two primary objectives in mind: efficient utilization of the radio spectrum and highly reliable communications.

      A CR system senses and learns about the outside environment in a circular process as shown in Figure 4. The outside world provides excitation, CR processes and analyzes the excitation to extract information that could help improve its system performance. This happens in the observation phase of a sensing circulation and input/output information is analyzed to get the capacity status. Based on such information, the CR system plans and allocates communication resources, makes a decision, initiates the access control process to start communication. Based on the information and learning rules, the system adjusts itself to optimum state before proceeding with the next round of sensing (observation) process.

2.7 CR Etiquette
What makes a CR network function orderly would be a whole package of action regulating spectrum etiquette. The etiquette, not easy to take shape though, would cover specifications and rules on RF spectrum, air interface, protocols, spatial-temporal model as well as high-level negotiation rules that targets at higher efficiency of spectrum use. The spectrum lease program, user priority policy and RKRL all help in the regulated spectrum sharing that is needed by CR.

      The spectrum lease program is a negotiation program similar to the handshake protocol. The user priority policy aims to improve efficiency of spectrum management and protect the orderliness of public communications and users are granted with different priorities. The current characteristics of spectrum allocation have defined a default priority for spectrum use. CR would bring significant changes to the current personal services.

2.8 Cognitive Networks (CNs)
As CR research advances, Motorola and Virginia Tech propose an idea of
"Cognitive Networks". The CN, as its name implies, has the sensing capability and sensing process to help plan, decides and take proper actions. This means CN can adjust its configuration to respond and adapt to operational and environment changes. A CN can maximize operator磗 capabilities. Such an intelligent CN comprises CR nodes, which are the core to the network.

Manuscript received: 2006-03-23

Author Introduce:

  Tian Feng  is currently studying towards a Ph.D. degree in wireless communications and network signals processing at Department of Signal and Information Processing, Nanjing University of Posts and Telecommunications.

Cheng Shilun is studying for his Doctor磗 degree at Department of Information and Telecommunication Engineering of Nanjing University of Posts and Telecommunications. His research directions are the radio communications and network signals disposal.

Yang Zhen is the president of Nanjing University of Posts and Telecommunications. He is also a professor and a tutor of doctor postgraduate. His research mainly includes wireless communications and network signals processing, speech processing and modern speech communications technologies, and information security technologies.