Go direct to the page content

AIoT and Mobile Networking

AIoT and Mobile Networking

1 Backscatter Communication

Most wireless IoT does not require a high data rate but it consumes a lot of energy.

Because supplying power is difficult job, Bluetooth or Wi-Fi is excessive specification for IoT devices.

  • Reuse the signal resources of existing Wi-Fi infrastructure.
  • Save energy more than 10,000 times by modulating backscattered signal of the existing ambient signal at the tag side.
  • Guarantee communication distance of several tens of meters and a transmission speed in Mbps.
Research Topic

Devising MAC Control

for when many heterogeneous devices communicate at the same time in Wi-Fi backscatter.

2 Federated Learning for IoT

Distributed training parameters from IoT devices have to be transferred to the central cloud server.

Considering the growth of mobile computing power, federated learning on wireless network becomes important research issue.

  • Wireless network topology changes dynamicallly. New nodes can enter, existing nodes quit anytime.
  • Quality of the communication of nodes can fluctuate by variable latency and message dropping between edge devices and central server.
Research Topic

Robust Federated Learning Protocol

We are now researching about a robust federated learning framework and protocol that copes with the drastic change in network topology and quality of communication.

Federated Learning

Federated learning is distributed training technique that central server makes global model to combine the parameters which is trained in multiple edge devices. It guarantees data privacy and reduces communication cost.