What if the world of the Internet of Things wasn't Gaussian?
Interference is constantly increasing in wireless communications because of the increase in the number of objects communicating. It is added to the usual receiver noise. However, it does not have the same statistical properties, which should have a significant impact on receiver design. Carried out with researchers from Aalborg, the first measurements of this interference in the 868 MHz band highlight its non-Gaussian nature.
5G and future networks will see an ever-increasing density of connected objects. As these objects are not coordinated, there is little opportunity to limit interference. It is therefore crucial to characterise interference in order to understand its impact on coding, waveform and receiver design. Although a number of theoretical models have been developed, there is very little experimental validation. This work partly fills this gap by performing a statistical analysis of recent measurements in the unlicensed band 863 - 870 MHz in different areas of Aalborg, Denmark (see Figure). In particular, the measured data show that the tail of the interference power distribution is heavy, confirming the predictions of theoretical models.
1. Interference is the consequence of simultaneous transmissions from different devices on the same channel as that observed by a given receiver. Since Middleton (1977), stochastic geometry has played an important role in characterising the statistical properties of interference. The interferents are located according to a homogeneous point Poisson process, and non-Gaussian interference distributions naturally appear. A significant result is that the behaviour of the tail of the distribution is dominated by the strongest interferent and that sub-exponential distributions often appear.
2. This work presents the first comparison of these theoretical results with experimental data. Measurements of received power were made at 5 different locations in Aalborg (Denmark): city centre, business park, hospital, industrial area and residential area. Measurements were made at street level using a radio network analyser equipped with an omnidirectional antenna over a 2-hour period. All RF activity on the antenna in the 868 MHz ISM band (863-870 MHz) was recorded.
3. The measurement data obtained at the five locations confirm the hypothesis of the sub-exponential nature of the interfering power distribution. The graph shows the survival function of the data and known distributions (exponential, denoted a=2, and sub-exponential, denoted a=1.4, 1.6 and 1.8). It clearly validates the heavy tail of interference in the context of Internet of Things (IoT) communications. Interference models are the key to designing efficient coding and decoding strategies as well as efficiently adapting channel access and network topology. As such, the measurement data suggests that there is a need to reconsider the usefulness of Gaussian models in designing networks for the IoT.
Clavier, Troels Pedersen, Ignacio Larrad, Mads Lauridsen, and Malcolm Egan, Experimental Evidence for Heavy Tailed Interference in the IoT, IEEE COMMUNICATIONS
LETTERS 25 (2021), no. 3, 692-695.
