This paper presents the first edge computing framework for RTI systems combining intermediate sensor fusion with model quantization. Traditional RTI approaches [1] rely on centralized architectures (450 ms latency), while our key innovations enable 73% faster edge processing (120 ms) through hybrid ResNet architecture compressed via 8-bit quantization and intermediate fusion of RTI attenuation maps and RGB features. Our work introduces an on-device processing pipeline implemented on low-power IoT devices---including Jetson Nano, Raspberry Pi, and ESP32---that enables real-time inference without cloud-based computation.

Wireless sensor networks, made so that objects and people can be found without devices, are an important part of our high-tech world. This poster aims to show how heterogeneous convolutional neural networks can be used to improve a radio tomographic imaging system that can find people indoors precisely. In addition to original algorithmic solutions, the system's advantages include using properly designed and integrated devices---radio probes---whose task is to emit Wi-Fi waves and measure the received signal strength. Thanks to the use of the two-stage approach, the sensitivity, resolution, and accuracy of imaging have increased. Furthermore, our solution works well for radio tomography and other types of tomography because it is easy to understand and can be used in many ways.

The aim of the research is to develop a system enabling effective and efficient tracking of people inside buildings using radio waves. The presented concept uses radio tomography imaging (RTI) as a passive analysis of radio wave interference as well as active connections with transmitting and receiving devices---mainly smartphones. A long short-term memory (LSTM) neural network was used to solve the inverse tomographic problem of converting measurements into images. The presented concept uses a proprietary design of transducers, which are transmitting and receiving devices that can exchange information with each other and establish connections with other devices. The novelty is the hybrid nature of the people location system, using both device-free and device-based methods. Another new approach is using the LSTM network to solve the inverse problem in RTI. Both solutions make the location system much more flexible, which makes imaging much more accurate and reliable.

Location systems and their daily use in industry and individual users have become an inseparable part of our everyday functioning. However, often they are used in the so-called intelligent buildings. Unfortunately, just like location methods operating outside buildings are precise enough to determine our location (e.g. GPS systems for drivers), the situation of locating people inside the building is different, especially when such people cannot be equipped with additional devices (e.g. a transmitter signal). Nevertheless, it is essential in communication routes, large clusters of people, such as airports, railway stations or office spaces. At the same time, determine the number of people in space and the development of services that do not require direct contact (automatic patient/client registration) in times of a possible pandemic.