Bilişsel Radyoda Özdeğer Tabanlı Spektrum Sezme Yöntemleri

Abstract

Kablosuz haberleşme sistemleri için gerekli olan frekans spektrumu doğal olarak sınırlı bir kaynak olduğundan etkin bir şekilde kullanılması büyük öneme sahiptir. Birçok ülkede mevcut spektrumun büyük bir kısmı tahsis edilmiş, spektrum kıtlığı problemi görülmeye başlanmıştır. Mevcut durumda kullanılan sabit spektrum erişimi tekniği ile, her servis için belli bir frekans bandı ayrılmıştır ve sadece lisanslı (birincil) kullanıcılar bu bandı kullanabilmektedir. Bu nedenle, frekans bandı boş olsa dahi lisanssız (ikincil) kullanıcıların bandı kullanmasına izin verilmemekte, dolayısıyla spektrum verimliliği düşük seviyelerde olmaktadır. Kullanıcı sayısı ve veri ihtiyacındaki hızlı artışla birlikte, spektrum verimliliğinin artırılması çok daha önemli hale gelmiştir. Bunu sağlamanın yolu, dinamik spektrum erişim teknikleri kullanmaktır. Bu tekniklere imkân veren bilişsel radyo, spektrum verimliliğini artırmak adına umut verici teknolojilerden biri olarak karşımıza çıkmaktadır. Bilişsel radyo, spektrumu sezebilir, boş frekans bandlarını tespit edebilir ve birincil kullanıcıların olmadığı zamanlarda bu boş bandların ikincil kullanıcılar tarafından kullanılmasına olanak sağlar. Lisanslı mevcut spektrumun lisanslı olmayan bir şekilde yeniden kullanımına olanak sağlayan bilişsel radyo, spektrum sezme ve ölçme teknikleriyle sınırlı olan mevcut band genişliğinin en etkin biçimde kullanılmasında önemli rol oynamaktadır. Bilişsel radyonun en önemli işlevlerinden biri spektrum sezmedir. Spektrum sezmede amaç, spektrumda periyodik olarak algılama yaparak lisanslı kullanıcıların hareketliliğini ve spektrumun durumunu saptamaktır. Bilişsel radyo alıcı-vericisi, kullanılmayan spektrumu ve lisanslı spektrum boşluklarını tespit ederek birincil kullanıcılara girişim yapmadan spektruma erişim yöntemlerini belirler. Literatürde spektrum sezme için enerji algılama, uyumlu süzgeç, çevrimsel durağan özellik algılama, özdeğer tabanlı algılama gibi birçok farklı teknik önerilmiştir. Bu tez çalışması kapsamında, çeşitli spektrum sezme teknikleri incelenerek, güçlü ve zayıf yönleriyle birlikte verilmiştir. Enerji algılama, çevrimsel durağan özellik algılama ve özdeğer tabanlı algılama için kullanılan farklı yöntemlerin MATLAB yardımı ile bilgisayar benzetimleri yapılmıştır. Özdeğer tabanlı spektrum sezme için, iteratif yöntemler (güç iterasyonu ve ters iterasyon) yoluyla hesaplanan özdeğerler kullanılarak, farklı iterasyon değerlerine göre algılama olasılığı eğrileri elde edilmiştir. Ayrıca, önerilen özdeğer farklarına dayalı yöntemin başarımı incelenerek, literatürdeki yöntemlerle karşılaştırması sunulmuştur.

The radio frequency spectrum which is required for wireless communication systems is a scarce natural resource and its efficient usage has a great importance. The spectrum is divided into different spectrum bands allocated to different services such as fixed, mobile, satellite and broadcast services. In many countries most of the current spectrum is already assigned and spectrum scarcity problem is encountered. It is a fundamental problem facing the future wireless systems to meet the demand for future services. Regulatory bodies in the world (including the Federal Communications Commission in the United States and Ofcom in the United Kingdom) found that most radio frequency spectrum is inefficiently utilized. Cellular network bands are overloaded in many countries, but other frequency bands such as military, amateur radio and paging frequencies have low spectrum occupancy. Independent studies performed in some countries confirmed that observation, and concluded that spectrum utilization depends on time and place. With the use of fixed spectrum access (FSA) method for spectrum allocation in wireless communication systems, each service has a dedicated frequency band and only licenced (primary) users have rights to use this band. Thus, even if the frequency band is empty for a while it can not be used by unlicenced users which leads to low spectrum occupancy. Increasing efficiency of the spectrum is an urgent need as the number of wireless users and demand for data are increasing rapidly. This can be achieved by dynamic spectrum usage. A promising technique to improve spectrum utilization that is the enabling technology for dynamic spectrum access is cognitive radio (CR). The concept of cognitive radio was first proposed by Joseph Mitola III in a seminar at KTH (the Royal Institute of Technology in Stockholm) in 1998 and published in an article by Mitola and Gerald Q. Maguire, Jr. in 1999. It was a novel approach in wireless communications. Cognitive radio is a form of wireless communication in which a transceiver can intelligently detect which communication channels are in use and which are not, and instantly move into vacant channels while avoiding occupied ones. This optimizes the use of available radio frequency (RF) spectrum while minimizing interference to other users. In its most basic form, cognitive radio is a hybrid technology involving software defined radio (SDR) as applied to spread spectrum communications. Cognitive radio has the ability of sensing the spectrum, detecting idle frequency bands and allowing unlicenced (secondary) users to use those bands when the primary users do not exist. Each cognitive radio user must sense and decide the available frequency bands of the spectrum, select the best available frequency band, coordinate access to this channel with other users and vacate the channel immediately when a licenced user is detected. These capabilities correspond the four main functions of a cognitive radio respectively: spectrum sensing, spectrum decision, spectrum sharing and spectrum mobility. Being the focus of this study, spectrum sensing by far is the most important component for the establishment of cognitive radio. Spectrum sensing is usually understood as measuring the spectral content, or measuring the radio frequency energy over the spectrum; but when cognitive radio is considered, it is a more general term that involves obtaining the spectrum usage characteristics across multiple dimensions such as time, space, frequency, and code. It also involves determining the types of signals occupying the spectrum including the information of modulation, waveform, bandwidth and carrier frequency. However, this requires more powerful signal analysis techniques with additional computational complexity. Spectrum sensing aims to detect the availability of spectrum holes by sensing the spectrum periodically. Spectrum hole is defined as a frequency band assigned to a primary user but that is vacant in a given place at a given time. Cognitive radio transceiver detects unused spectrum and spectrum holes and then decides spectrum access techniques without causing any interference to primary users. The availability of spectrum holes can be estimated with different spectrum sensing techniques. These spectrum sensing techniques can be classified into three groups generally: primary transmitter detection, primary receiver detection and interference temperature management. Primary transmitter detection is based on the detection of a weak signal from a primary user transmitter by exploiting the local observations of cognitive radio users. Another method for detecting spectrum holes is to detect the primary users which are receiving data within the communication range of a cognitive radio user. Currently, this method is only feasible for sensing TV receivers. Third method is the interference temperature management model that limits the interference at the receiver through an interference temperature limit. This limit means the amount of new interference the receiver could tolerate. Cognitive radio users can use the spectrum band if they do not exceed this limit. Although this model is the best fit for the objective of spectrum sensing, the difficulty of this model is determining the interference temperature limit accurately. In this study, we consider primary transmitter detection methods. There are different common spectrum sensing techniques proposed in the cognitive radio literature for spectrum sensing such as energy detection, matched filtering, cyclostationary detection and eigenvalue based detection. In this study, these different techniques are investigated and their powerful and weak aspects are given. Energy detection, also known as radiometry or periodogram method, is the most common way of spectrum sensing because of its low computational and implementation complexities beside not needing any information about primary user signal. Some of the challenges with energy detector based spectrum sensing technique are selection of the threshold for detecting primary user signals (noise uncertainty problem), inability to differentiate interference from primary users and noise while secondary user is transmitting, and poor performance under low signal to noise ratio (SNR) values. Since the threshold used in energy detection algorithm depends on the noise variance, a small noise power estimation error causes significant performance degradation. Matched filtering is known as the optimum method for detection of primary users when the transmitted signal is known. The main advantage of matched filtering is the short time to achieve a certain probability of false alarm. Another approach is cyclostationary based spectrum sensing. Cyclostationary feature detection is a method for detecting primary user transmissions by exploiting the cyclostationarity features of the received signals. These features are caused by the periodicity in the signal statistics like mean and autocorrelation or they can be intentionally induced to assist spectrum sensing. Eigenvalue based spectrum sensing techniques exploit the covariance matrix of received signals at secondary users. These methods overcome the noise uncertainty problem that is encountered in energy detection method, and does not require any knowledge of signal, channel and noise power. In the concept of this thesis, computer simulations are made for energy detection, cyclostationary based and eigenvalue based spectrum sensing algorithms in MATLAB. For the eigenvalue based spectrum sensing, different techniques given in the literature namely EME (Energy with Minimum Eigenvalue), MME (Maximum/Minimum Eigenvalue), GLRT (Generalized Likelihood Ratio Test) and RLRT (Roy’s Largest Root Test) are simulated. Probability of detection curves are given for the methods according to SNR values and different noise samples values. For the MME algorithm, some iterative solutions (power iteration and inverse iteration) in the linear algebra are used to calculate maximum and minimum eigenvalues of covariance matrix, and probability of detection curves are given by using the calculated eigenvalues for different iteration numbers. Moreover, a new method for eigenvalue based spectrum sensing is proposed which is based on the sum of the squares of eigenvalue differences. Performance comparisons are made for the proposed method with the other eigenvalue based spectrum sensing methods in the literature. Eigenvalue based methods (including the proposed method) overcome the noise uncertainty problem, which can even perform better than the ideal energy detection when the signals to be detected are highly correlated. The methods can be used for various signal detection applications without requiring the knowledge of signal, channel and noise power.