Tutorial-1: Artificial Intelligence Augmented Robotic Neurorehabilitation

Shyamanta M. Hazarika
Biomimetic Robotics and AI Lab, Mechanical Engineering,
Indian Institute of Technology Guwahati

Cota Navin Gupta
Neural Engineering Lab, Bioscience and Bioengineering
Indian Institute of Technology Guwahati

Technological advances, particularly in Artificial Intelligence (AI) and Robotics ispushing the boundaries of Robotic Neurorehabilitation. Progress in Robotics, AI andMachine Learning (ML) combined with insights into Neuroscience are definingnovel strategies for motor therapy driving effective neurorehabilitation. Further,wearable robotics, virtual reality, brain machine interfaces and bionics are playing aleading role. Neurorehabilitation robots require sophisticated control algorithms toensure 'assistance-as-needed' and 'retention of residual skills'. This calls formethods for recognition of human intent and residual capabilities from bio-signals,particularly electroencephalogram (EEG) and electromyogram (EMG). The tutorialwould provide an overview of AI techniques for neurorehabilitation and highlighthow it is driving therapies in the emerging area of robotic neurorehabilitation. Brain machine interfaces (BMIs) for neurorehabilitation are often accomplished using EEGsignals, local brain signals generated through motor imagery (MI) i.e., mentalimagination of a particular task. MI have been shown to lead to mental fatigue (MF)and deterioration of EEG. The tutorial will include a self-contained introduction toMI EEG; MI and MF correlation and introduce adaption of feature extraction for MIBMIs. Starting from the dominant paradigm of assistive control, the tutorial wouldgo on to include collaborative cognitive control that adjust assistance based on notonly the patient's performance but also physiological responses. The tutorial is notonly intended to serve as an introductory course but also discuss challenges andopportunities, highlighting emerging trends within this area and give pointers todirections of future research.

Tutorial-2: Extraction of Excitation Component Information from Speech and its Applications: A Review

Sudarsana Reddy Kadiri, Paavo Alku
Department of Signal Processing and Acoustics, Aalto University, Finland

B. Yegnanarayana
Speech Processing Laboratory, IIIT-Hyderabad, India

The objective of this tutorial presentation is to give an overview of the techniques for derivingexcitation component information of speech production, which can be attributed mostly to thevoice source at the glottis and to present significance of excitation for the analysis of variousvoices.
Humans have an extraordinary capability of producing a wide variety of variations in voice by manipulating the complex interaction between the voice source (the glottal volume velocitywaveform generated by the vocal folds) and the vocal tract system. This flexibility allows thespeaker to produce voices ranging from verbal sounds to nonverbal sounds including differenttypes of voice qualities, vocal emotions and singing styles. The excitation component plays themost important role in generation of all these variations of voice. Moreover, the excitation componentconveys useful complementary information to the existing widely used features such asmel-frequency cepstral coefficients (MFCC) and perceptual linear prediction (PLP) coefficients,which characterize mainly the response of the vocal tract system. The proposed tutorial presentationgives a comprehensive overview of techniques for deriving excitation component informationfrom speech signal and its applications. The challenges in deriving the excitation component information,and some possible avenues for exploring the recent deep learning techniques will alsobe discussed. The tutorial is self-contained and consists of four parts. In order to appreciate the sophisticationof voice source in speech production, Part I gives the physiology of speech production bydescribing the organs that are involved in production of speech. Part II describes the excitationinformation that can be obtained from speech and from non-acoustic measurements. The significanceof the excitation component in the analysis of different types of voices is discussed in PartIII. Finally, Part IV gives the main contributions/findings, challenges, and possible future avenues.

Tutorial-3: Inter-antenna interaction and its effects in MIMO wireless

Rakhesh Singh Kshetrimayum
Department of Electronics and Electrical Engineering,
Indian Institute of Technology Guwahati, Assam

It is a well-established fact that Multiple-input-multiple-output (MIMO) wireless is already employed in 5G wireless and it is a technology expected to be used in beyond 5G and 6G wireless. The main advantages of the MIMO wireless in comparison to the conventional Single-input single-output (SISO) wireless are its two major gains viz. diversity gain (DG) and multiplexing gain (MG). The MG can be derived from the channel capacity of the MIMO system. But due to utilization of multiple antennas at the transmitter and receiver, the problem of Inter-antenna Interaction (IAI) is inevitable in the MIMO wireless. The two most important IAI parameters for MIMO antennas are envelope correlation coefficient (ECC) and antenna correlation coefficient (ACC). The ECC can be accurately calculated from the 3-D radiation patterns of the MIMO antenna. From the ECC, ACC can be obtained. Once ACC matrix of the MIMO antenna is available, it can be used to observe the effect of MIMO antenna on the MIMO system's performance such as the DG and the MG. It can be also utilized for the calculating the channel capacity loss (CCL). The CCL for single-user MIMO (SU-MIMO) and multi-user MIMO (MU-MIMO) will be calculated and investigated for some typical point-to-point MIMO and distributed MIMO systems respectively. It has been apparent that the ECC for MIMO antenna is the most important IAI parameter which decide the performance of the MIMO antenna in the MIMO systems. So the aim and objective of MIMO antenna engineers should be minimizing the ECC. In wireless domain, IAI has been completely eradicated by techniques such as Spatial modulation. A brief discussion on such techniques will be initiated. Several MIMO antenna examples will be provided which minimize mutual coupling and the ECC to a great extent and study their effect on the MG and the DG will be carried out. A brief discussion on the evolution of the MIMO systems, from SU-MIMO to MU-MIMO, MU-MIMO to massive MIMO (mMIMO), mMIMO to cell-free massive MIMO (CF-mMIMO) will be pursued in this talk.

Tutorial-4: Photonics and plasmonics for intrachip and chip-to-chip Interconnects

P. K. Basu
Retired Professor, Institute of Radio Physics and Electronics
University of Calcutta

In spite of the remarkable development of transistors with 3-nm node, the network-on-chip (NOC) involving complex circuits faces the problem of high power dissipation due to interconnect bottleneck, degrading the performance of data centres and high performance computers. Photonics, particularly on the silicon platform, offers viable solution to the interconnect problem and electronic-photonic integration on Si is currently a hot topic for R&D activities and industrial applications. The poor emission from Si requires hybrid integration with III-V compound and alloy based lasers and encourages intensive research on GeSn alloys for lasers and other photonic components. The ultimate size of photonic devices is however diffraction limited, which allows individual devices to have size of light wavelength in the micrometer range. Plasmonics, in particular surface plasmonics, allows realization of sub wavelength sized devices with ultra-low power consumption at high bit rate and therefore seems to be the ultimate solutions for NOC. The very small propagation length of plasmonic waves can be overcome by Surface Plasmon Amplification by Stimulated Emission of Radiation (SPASER) replacing photonic nanolasers used in communication and networking. The area is still in its infancy, but is subject of current intense research.The proposed tutorial aims to cover the basics of Si-Photonics and Plasmonics, and the application areas mentioned, and to present current state-of the-art technology.