Spectrum occupancy investigation: Measurements in South Africa

Abstract

Regulatory bodies have predicted an impending shortage in commercial radio frequency spectrum in the near future.However, due to outdated regulatory practices, many of these bands are in fact inefficiently underutilised. Spectrum measurement campaigns have be carried out around the world to determine the extent to which this is true. However, there still seems to be a lack of knowledge regarding spectrum occupancy in South Africa. A spectrum measurement system was thus designed and employed to measure the spectrum occupancy of the ultra-high frequency (UHF), global system for mobile communications (GSM) 900 MHz and GSM 1800 MHz bands through a measurement campaign in the Hatfield area of Pretoria, South Africa. A method for determining spectrum occupancy, from raw spectrum measurements, has been described and used to calculate the average spectrum occupancy of these bands. Occupancy in the UHF band was found to be fairly constant at approximately only 20%. While the maximum occupancy of the GSM 900 MHz band was calculated to be much higher at approximately 92% and that of the GSM 1800 band to be approximately 40%. However, the GSM 900 MHz and 1800 MHz bands did exhibit fluctuations in occupancy of between 10% and 20% respectively according to the time of day. Slight variations in occupancy of between 1% and 3% were also evident over the days of the week. These results are placed into context by a comparison with the findings of various other measurement campaigns from around the world. When compared, occupancy was generally found to be lower in the UHF bands but higher in the mobile bands.

Spectral occupation of TV broadcast bands: Measurement and analysis

“…A wideband measurement system, described in [5], was used to measure the power spectrum of the broadcast TV bands at six different locations, labelled A to F, on the Hatfield campus of the UP, South Africa. Both the VHF (174-254 MHz) and UHF bands (470-854 MHz) were considered.…”

“…The thresholds were thus chosen as λ = −90 dBm (55 dBµV/m) and λ = −80 dBm (65 dBµV/m) for the VHF and UHF bands respectively. The second approach was based on the maximum normal fit (MNF) algorithm, proposed in our previous work [5]. In this approach the noise distribution is separated from the signal distribution of a measured set of channels and λ is calculated as the point at which they intersect.…”

“…The campaign sought to quantify the effective usage of the TV broadcast bands and was motivated by the need to identify possible portions of unused spectrum for use by secondary users. Spectral opportunities are compared using three different mechanisms: Localised spectrum measurements obtained by using a specific hardware platform [5], spectrum assignments done by the regulator [3] and a locally available static GLDB that is derived from standard wireless propagation models [6].…”

Cooperative Prediction for Cognitive Radio Networks

“…A statistical approach to calculating the noise floor, called the (MNF) method, may be employed to calculate λ i [5]. The MNF method has been adopted to perform SS in this article and will be further investigated in Section 4.…”

“…Firstly, a practical approach to implementing the maximum normal fit (MNF) algorithm for SS, a statistical approach to calculating the noise floor [5], is investigated. Secondly, the prediction accuracies of various prediction methods are compared under different traffic conditions.…”