Dr Manijeh Bashar,
University of Surrey
Dr Kanapathippillai Cumanan,
Prof. Alister Burr
University of York
£50 Tutorial only | Tutorial + Full Conference access from £120
Half-day Tutorial, Monday 31st August 9:00 (BST)
Further pricing details can be found on the registration page.
In cell-free massive MIMO, distributed access points (APs) are connected to a central processing unit (CPU) and jointly serve distributed users. This approach provides many of the benefits of cloud radio access network (C-RAN – which provides a clear means to implement it), enjoying lower average path loss as well as distributed signal processing. As a result, by removing the concept of ‘cell’ and ‘cell-edge users’, Cell-Free Massive MIMO is a promising system model as it provides all wireless devices with a great data rate service.
One of the most practical challenges in cell-free massive MIMO is the connection and internal communication between APs. The “radio stripe” system is a low-complexity and cost-efficient technology to implement cell free massive MIMO. The link between the APs and the CPUs are called fronthaul links, while backhaul links are exploited for the connections between the CPUs.
this tutorial will provide an overview of a complete and practical cell-free massive MIMO system, including comprehensive signal processing schemes, mathematical and statistical analyses for the fronthaul links, energy-efficient resource allocation, and user assignment techniques.
This tutorial will consider several significant challenges with a view to contributing to 5G and beyond.
The main objectives are to address:
- Optimal fronthaul for energy-efficient green communication
- Effect of the capacity limitations of fronthaul links
- Machine learning techniques to improve the performance
Structure and content
The first part of the tutorial introduces the basic concepts of cell-free massive MIMO. Different ways to implement massive MIMO are described. Next, the basic principle of the uniform quantizer is reviewed, including finding the optimal step size for the uniform quantizer. The achievable rate is next explained for the case of unknown channel gain at the receiver.
We then present an efficient user assignment algorithm and show that this achieves a further improvement.
In the next section of the tutorial, the total energy consumption in a cell-free massive MIMO uplink is modelled, enabling us to define the energy efficiency optimization problem in cell-free massive MIMO uplink.
We then explain how to train a deep convolutional neural network (DCNN) in an off-line manner to determine the power control coefficients based on the long-term statistics (LTS) to maximize the data rate and energy efficiency.
Finally, we discuss practical implementation.
The structure of the tutorial is given as:
- Fundamental concepts of cell-free massive MIMO
- Basic architecture of system; pilots and pilots assignment; conjugate beamforming/MRC, ZF and MMSE; determining capacity; fronthaul problem
- Fronthaul quantization and optimization
- Principles of quantization, Bussgang decomposition, optimal quantizer, vector quantization
- System optimization
- Max-min rate optimization approaches and results, energy efficiency optimization and results, machine learning techniques to improve the performance, user assignment algorithm
- Implementation of cell-free massive MIMO
- Fronthaul alternatives, “Radio Stripes” system; relationship with C-RAN, different PHY splits