Occupation Measurements Supporting Dynamic Spectrum Allocation for Cognitive Radio Design

Abstract

Abstract-Efficient use of radio spectrum is a necessity for 3000 MHz were intended to clarify, among other aspects, future wireless systems. This can, for example, be accomplished the questions of channel occupation, how those channels are by cognitive radios. An important point for designing future sys-occupied (i.e., modulation scheme, data rate, etc.), and how tems is to gain knowledge about the occupation of the frequency .'. ' ' bands which may be used by the cognitive radio. This includes many tnsitr arept in a and andtwhre teyiae the general characterization of the traffic density at the frequency located. In [6] a concept of spectral agile software defined bands as well as a detailed analysis of the temporal frequency radios has been discussed where a method for estimating occupation. Thus, this paper describes measurements taken in underutilized spectrum bands has been presented. the GSM network at three different scenarios. Furthermore, INTRODUCTION system. Therefore, the major incentive of this paper is to T HE demand for mobile services steadily increases. Often present measurements of a typical primary system, thereupon radio resources are proclaimed to be scarce resources, but describing the statistical characteristics and discussing potenthis scarcity only arises due to an inefficient usage. By em-tial consequences for the design of cognitive radios. ploying adaptive allocation and dynamic pricing mechanismsThe remainder of the paper is organized as follows: In these resources can be exploited more efficiently. One possible section II the measurement system is described by specifying technique to realize dynamic spectrum access may be based the equipment used and the parameters chosen. The scenarios on spectrum pooling [1], [2].in which the measurements have been performed are charIn a spectrum pooling system the same spectrum is locally acterized in section III. Measurements provide information used by secondary users following proper rules to transmit about the time-frequency occupation of the frequency range wirelessly without disturbing the incumbent user. For example, under consideration. This time-frequency occupation serves spectrum pooling systems may be realized by primary systems as starting point for the statistical analysis conducted in allowing a secondary system to send in the time-frequency section IV. Possible impacts of the results on the design of gaps whenever the primary users do not transmit. Thus, the cognitive radios are discussed in section V. Finally, the paper secondary system needs to detect the primary system and it is summarized. must be able to decide about transmitting and adapting its radio access technology to the current spectrum configura uoon.rTion. eAuREment sYstAssuming that the secondary system gets no signaling inforThe measurement equipment mainly consists of an antenna, mation from the primary system, the secondary system needs a spectrum analyzer and a computer (see figure 1). The to learn about the past and act accordingly. This requirement intention ...

A Survey of Measurement-Based Spectrum Occupancy Modeling for Cognitive Radios

“…In [20], another campaign in Atlanta, USA, was performed where the power spectral density results were given. In [21], [22], [23] and [24], some specialized systems, such as Wi-Fi, GSM and public safety systems, were studied, where measurements were taken to show that the Wi-Fi band was still relatively vacant while the public safety band was relatively full, at the time of study. A more important conclusion from [23] is that there are seasonal, weekly and daily trends in the spectrum occupancy statistics.…”

“…The calculated PDF for the GSM900 channel rises slowly then drops quickly when the power level increases, while the calculated PDF for the GSM1800 channel rises quickly and then drops slowly when the power level increases. In [22], another measurement campaign for the GSM system was performed for different traffic channels. The obtained PDF showed similar asymmetry to [34], but interestingly, some traffic channels in some scenarios show double-peaked PDFs.…”

“…In [18], the detection threshold was set 6 dB above the minimum power level, not the noise floor as in (3). In [22], the noise samples were assumed Gaussian such that the margin M (f ) was chosen as a function of the standard deviation to achieve a probability of false alarm of 0.003. There are also other ways of setting the detection threshold, borrowed from other research areas, such as Otsu's algorithm, recursive and adaptive thresholding [30].…”

Spectrum Usage in Cognitive Radio Networks: From Field Measurements to Empirical Models

“…3) [38]. In some other channels, the load variation pattern is characterized by a highly random component resulting from a significant number of random factors such as the number of incoming and outgoing users, resource management policies, social behavior of users, external events and so forth, which may lead to highly unpredictable variations (e.g., see [39]). In such a case, a stochastic modeling becomes a more convenient approach [38].…”

Analysis of Spectrum Occupancy Using Machine Learning Algorithms

“…For instance, the statistical and spectral occupation analysis of the measurements was presented in [2] in order to study the traffic density in all frequency bands. In [3], auto-regressive model was used to predict the radio resource availability using occupancy measurements in order to achieve uninterrupted transmission of secondary users.…”