Ultra-wideband

The term ultra-wideband refers to the radio communication with a large effective bandwidth (between 20-25% of the center frequency depending on the definition) of ≥500MHz. This large bandwidth gives the UWB the possibility to transmit a large amount of data at a short distance. It is also considered as a very secure method of communication since the signals can be designed to appear as imperceptible random noise to conventional radios. It also operates at low power since it works by impulse modulation as opposed to the frequency modulation of narrowband communications. Impulse modulation means that there is no need for a carrier wave. Figure 1 shows the difference between the narrowband and the ultra-wideband. The impulse modulation, as opposed to amplitude, frequency or phase modulation, is also non-interfering. This is due to the high-time resolution. It results in a method insensible to the well-known multipath propagation problem. UWB usually relies on many rapid short pulses that occupy the entire bandwidth. This characteristic enables high spatial resolution when UWB is used as radars (up to a few millimeters).

Ultra-wideband radars (UWB) have been the subject of many promising works on various related fields such as heart rate monitoring, sleep monitoring, occupancy detection, 3D imaging or fall detection. Slow-time information are usually used with a Fast Fourrier Transform (FFT) to infer a patient heart rate, though several other methods have been used to improve the accuracy. This method is a way to exploit the manifestation of the Doppler effect and pickup small body motion that can also be applied for sleep monitoring. Principal Component Analysis can be used on the UWB data to suppress clutter and detect the occupancy of a room. Kirchhoff Migration and Stolt frequency-wavenumber (F-K) Migration, used in imaging, can be used as well with UWB data offering a fast, accessible and cheap alternative for security imaging devices. Additionally, interesting features can be extracted from the micro-Doppler effect which can then be utilized for many classification methods for fall detection. Nevertheless, most of the work found in the literature exploit only one radar in a laboratory setting. In our knowledge, there are no realistic deployment of a plurality of radars for the purpose of activity recognition in a realistic home environment.

Our research team has access to a real-size smart home prototype and conducts experimental research inside the University. This smart home is an infrastructure comprising a living room, a kitchen, a bedroom, and a bathroom. The whole environment is about 40 square meters. There are student desks all around the infrastructure with windows to see what’s going on inside. Our team is sensors agnostic and, therefore, there are various types of technologies in the prototyping smart home including motion, ultrasonic, temperature, RFID, power analyzer, BLE beacons, and other types of sensors. The reason is, like many other researchers in the community, our team has not yet found the perfect solution to solve the long going problems regarding the tracking and the recognition of the ongoing ADLs in the smart home. The challenge of ADLs recognition in smart homes is to create an Artificial Intelligence (AI) on the top of the smart home’s sensors, offering only a partial vision of the environment, and to deduce the actions of an unaware person inside. The difficulty resides not only in the limited information but also in the almost unlimited courses of actions humans can do in their environment. Hundreds of papers have been published on this topic over the last decades. Yet, it eludes us because we need to solve it with little intrusiveness, low cost, and with good overall precision. The UWB radar technology, which was presented in Section 2, has the potential to partially address these challenges. In particular, the XeThru X4M200, from Novelda, could be a cost-effective solution at $249 US per SoC if a few of them could cover an apartment and enable ADLs recognition as a standalone technology.