In this project, the channel code design problem for massive Internet of Things (IoT) is considered by taking into account the unique characteristic of massive IoT, i.e., short length and bursty nature. Our aim is to develop practical coding strategies for both contention-based and grant-free access schemes.
In the near future our everyday physical objects will be transformed to information sources, communicating with each other and the underlying data transport infrastructure. This will create an ecosystem of connected devices and revolutionize the way we live today and open new roads for creativity and innovations. Around 50 billion devices will be installed by 2020 and they will generate an enormous data traffic. Providing connectivity and handling such a traffic are beyond the capabilities of current wireless standards, calling for breakthrough innovations in communication strategies. This project develops novel communication strategies for future wireless systems to support a large number of devices and diverse service requirements. Current wireless standards have been designed and optimized for human-based traffic, that is long packets are used for sending the data. However, in massive IoT, devices usually generate small packets, therefore using the existing standards would be wasteful of resources. More specifically, new channel coding techniques should be designed for small packets in massive IoT applications.In contention based random access, the transmission phase consists of allocating individual users to
dedicated, orthogonal resource blocks. The codeword length in massive IoT is assumed to be only
few kbs and the objective is then to design short block length codes that achieve high rate efficiency in the given time and frequency resources. To meet this requirement structured (algebraic) low-density parity-check codes will be developed. We will also develop a multi-level Raptor code, where multiple devices are transmitting at the same data channel using the same Raptor code and degree distribution function, but with different power levels. The received signal at the BS can then be realized as the coded symbols of a superimposed multi-layer Raptor code, where a multi-stage decoder and SIC are used to decode the devices' messages. The new work required here includes a degree distribution design for multi-layer Raptor codes with short message lengths, as the current analysis and design of Raptor codes are for asymptotic case when the message length goes to infinity and designing an optimal power control strategy to maximize the throughput and the number of devices which can be supported in a given number of resource blocks.
This program is with collaboration of Professor Mischa Dohler, King's College London, and Dr Gianluigi Liva, German Aerospace Centre.
The opportunity ID for this research opportunity is 2410