Doshisha University International Science and Technology Course

  • S
  • M
  • L
  1. Home
  2. Programs & Courses
  3. Graduate School of Science and Engineering : Information and Computer Science (Laboratories : Laboratory for Information Theory and its Applications)

Graduate School of Science and Engineering
Information and Computer Science

Laboratory for Information Theory and its Applications

Website of the Laboratory 【In Japanese】

We invite you to join us to do research and development on high-speed, reliable, robust and secure communications

Staff

JunCheng

Jun CHENG
[Professor]
Acceptable course
Master's degree course
Doctoral degree course
Telephone : +81-774-65-6295
jcheng@mail.doshisha.ac.jp
Office : YE-215
Database of Researchers
No Image

Tomotaka KIMURA
[Assistant Professor]
Acceptable course
Master's degree course×
Doctoral degree course×
Telephone : +81-774-65-6294
tomkimur@mail.doshisha.ac.jp
Office : YE-214
Database of Researchers

Research Topics

1.Wireless communications for IoT
2.Machine Learning for Wireless Communications
3.Coding for multiple-access channel
4.Coding for MIMO systems
5.Adaptive signal processing for wireless communications
6.Channel coding and wireless communications
7.Robust wireless netwrorking
8.Communication network analysis
9.Network architecture design

Research Contents

Information theory, coding theory, communication theory, teletraffic theory and cryptography are fundamentals in constructing modern information network. Based on these basic theories, we do research and development on information transmission, with particular attention to reliable communications with channel coding and robust communications with teletraffic theory. We expect this focus to bring new and important scientific and applied knowledge to information transmission.
img1
Wireless Communications for IoT
The Internet of Things (IoT) bring wireless connectivity to anything, from tiny static sensors, machines to robots, vehicles and drones. A successful implementation of IoT calls for wireless technologies that are able to support a much larger number of connected devices, and that are able to fulfill much more stringent requirements on latency and reliability than current systems. A generic scenario in IoT involves a very large number of idle (inactive) devices, but in a typical application, only a small (unknown) subset of devices are active at any given instant. In this scenario, short packet are the typical form of the traffic with the need of low latency. The classical information-theoretic results of a K-user multiple-access, however, are not applicable, since the assumptions of sustained connectivity and sufficiently large lengths of packets are not valid.
This research will develop and put forward a random-access strategy with short-packet transmission to meet the new requirements of IoT networks. Random access can support a very large number of devices to efficiently communicate in a sporadic and uncoordinated way. Some recent advances from information theory, e.g., the finite-length random coding bounds for the multiple access channel, provides us new information-theoretic metrics for short-packet transmission. Such leverage tools from coding theory as code on sparse graph, extrinsic information transfer charts, and message passing decoding will be utilized for the design of the novel multiple access paradigm and its analyses.
img2


An example of IoT network with a very larger number of devices, e.g., sensors for fire detection, vehicles, drones, robots etc..

Machine Learning for Wireless Communications
Recently, machine learning demonstrates a great advantage in solving optimization problems. A prominent advantage is its pure data-driven method, where the networks/systems are optimized over a large training data set, while the conventional model-driven method attempts to capture knowledge and derive decisions through explicit representation and rules in mathematics.
This project will focus on the application of machine learning to wireless communication systems, especially to signal detection and channel decoding over a class of uncertain/variable wireless channels. Machine learning in wireless communication systems has an advantage that it is easy, in computer simulations, to collect the large training dada set with automatic classification labelling of target data. Since the conventional probability models of the wireless channels have a gap between these models and real-world physics, the data-driven machine learning is potential way to close the gap, and thus enhance the robustness of wireless communications systems.
img3


An example of application of machine learning to wireless communications systems

Channel coding and decoding algorithms for wireless communication systems
In wireless communication systems, channel noise and interference would result in the unreliable information transmission. The interference, for example, in multiuser communication systems, is due to that multiple users share a common channel. In multiple-input multiple-output (MIMO) communication systems, the interference occurs due to simultaneous transmission from multiple antennas. Shannon’s coding theorem shows that the reliable information transmission is possible if the transmission rate is less than the channel capacity. We focus on the channel coding to realize the reliable information transmission in multiuser communication systems and MIMO systems. We aims high-reliable coding schemes and low-complexity decoding algorithms with the transmission rate approaching the theoretic limit. We will publish the results in leading international journals and conferences, and will transfer results from recent research in these areas to industry in mobile communications, wireless LAN, IoT, and satellite communications.
img4


An example of multiple-access communications in mobile networks

Robust wireless networking
In general, wireless communications are inferior to wired communications (e.g., optical fiber communications) in terms of stability. We thus study robust and reliable networking technologies to overcome various failures that occur in wireless networks. Especially, we consider sparse mobile ad-hoc networks, where the node density is very sparse. In conventional wireless networks, relay nodes use Store-and-Forward transmission to forward information. Specifically, when a relay node receives a packet from an adjacent node, the relay node forwards the packet as soon as possible. However, in sparse mobile ad-hoc networks, this method cannot be adopted because there are no neighbor nodes for most of the time. Therefore, Store-Carry-Forward (SCF) routing has been proposed for delivering messages. In SCF routing, when a node receives or generates messages, it stores them in its buffer. The node then carries the messages until it encounters other nodes. When this happens, it forwards the messages to the encountered nodes. By repeating this procedure, the messages are finally delivered to the destination node. So far, we have analyzed the performance of the existing SCF routing schemes, proposed some efficient SCF routing schemes, and realized various network services using SCF routing schemes.
img5

An example of sparse mobile ad-hoc networks

Keywords

  • Wireless communications
  • Internet of Things
  • Machine Learning
  • Error-correcting code


  • Signal processing
  • Communication network analysis
  • Network architecture design
  • Robust Wireless Networking