Author: Linga Cenkeramaddi

Video Content Analysis, Computer Vision, Pattern Recognition, Machine Learning, Deep Learning (VIGIL)

https://people.iith.ac.in/ckm/

N. S. Chilamkurthy, O. J. Pandey, A. Ghosh, L. R. Cenkeramaddi and H. N. Dai “Low-Power Wide-Area Networks: A Broad Overview of its Different Aspects,” has been accepted for publication in the IEEE Access Journal (2022).

Keywords:Low-power wide area networks,Internet of Things,Wireless communication, Costs,Wireless sensor networks,Market opportunities,Consumer electronics

Abstract:Low-power wide-area networks (LPWANs) are gaining popularity in the research community due to their low power consumption, low cost, and wide geographical coverage. LPWAN technologies complement and outperform short-range and traditional cellular wireless technologies in a variety of applications, including smart city development, machine-to-machine (M2M) communications, healthcare, intelligent transportation, industrial applications, climate-smart agriculture, and asset tracking. This review paper discusses the design objectives and the methodologies used by LPWAN to provide extensive coverage for low-power devices. We also explore how the presented LPWAN architecture employs various topologies such as star and mesh. We examine many current and emerging LPWAN technologies, as well as their system architectures and standards, and evaluate their ability to meet each design objective. In addition, the possible coexistence of LPWAN with other technologies, combining the best attributes to provide an optimum solution is also explored and reported in the current overview. Following that, a comparison of various LPWAN technologies is performed and their market opportunities are also investigated. Furthermore, an analysis of various LPWAN use cases is performed, highlighting their benefits and drawbacks. This aids in the selection of the best LPWAN technology for various applications. Before concluding the work, the open research issues, and challenges in designing LPWAN are presented.

More details: DOI: 10.1109/ACCESS.2022.3196182

A. Dayal, L. R. Cenkeramaddi, and A. Jha, “Reward Criteria Impact on the Performance of Reinforcement Learning Agent for Autonomous Navigation” has been accepted for publication in Elsevier Applied Soft Computing Journal (2022).

Key words:Deep reinforcement learning,Reward criteria,Autonomous navigation,Machine learning and artificial intelligence

Abstract:In reinforcement learning, an agent takes action at every time step (follows a policy) in an environment to maximize the expected cumulative reward. Therefore, the shaping of a reward function plays a crucial role in an agent’s learning. Designing an optimal reward function is not a trivial task. In this article, we propose a reward criterion using which we develop different reward functions. The reward criterion chosen is based on the percentage of positive and negative rewards received by an agent. This reward criteria further gives rise to three different classes, ‘Balanced Class,’ ‘Skewed Positive Class,’ and ‘Skewed Negative Class.’ We train a Deep Q-Network agent on a point-goal based navigation task using the different reward classes. We also compare the performance of the proposed classes with a benchmark class. Based on the experiments, the skewed negative class outperforms the benchmark class by achieving very less variance. On the other hand, the benchmark class converges relatively faster than the skewed negative class.

R. Jaiswal, S. R. Yeduri and L. R. Cenkeramaddi, “Single-Channel Speech Enhancement Using Implicit Wiener Filter for High-Quality Speech Communication” has been accepted for publication in the Springer International Journal of Speech Technology (2022).

Keywords: Edge computing, Non-stationary noise ,Raspberry Pi ,Spectral subtraction, Speech analysis,Stationary Noise ,Wiener filtering

Abstract:Speech enables easy human-to-human communication as well as human-to-machine interaction. However, the quality of speech degrades due to background noise in the environment, such as drone noise embedded in speech during search and rescue operations. Similarly, helicopter noise, airplane noise, and station noise reduce the quality of speech. Speech enhancement algorithms reduce background noise, resulting in a crystal clear and noise-free conversation. For many applications, it is also necessary to process these noisy speech signals at the edge node level. Thus, we propose implicit Wiener filter-based algorithm for speech enhancement using edge computing system. In the proposed algorithm, a first order recursive equation is used to estimate the noise. The performance of the proposed algorithm is evaluated for two speech utterances, one uttered by a male speaker and the other by a female speaker. Both utterances are degraded by different types of non-stationary noises such as exhibition, station, drone, helicopter, airplane, and white Gaussian stationary noise with different signal-to-noise ratios. Further, we compare the performance of the proposed speech enhancement algorithm with the conventional spectral subtraction algorithm. Performance evaluations using objective speech quality measures demonstrate that the proposed speech enhancement algorithm outperforms the spectral subtraction algorithm in estimating the clean speech from the noisy speech. Finally, we implement the proposed speech enhancement algorithm, in addition to the spectral subtraction algorithm, on the Raspberry Pi 4 Model B, which is a low power edge computing device.

More details: https://hdl.handle.net/11250/3013314

David Gonz´alez Le´on, Jade Groli, Sreenivasa Reddy Yeduri, Daniel Rossier, Romuald Mosqueron, Om Jee Pandey, and Linga Reddy Cenkeramaddi, “Video Hand Gestures Recognition using Depth Camera and Lightweight CNN,” has been accepted for publication in the IEEE Sensors Journal (2022).

Keywords:Gesture recognition,Cameras,Convolutional neural networks,Feature extraction, Streaming media,Hidden Markov models,Video sequences

Abstract:Hand gestures are a well-known and intuitive method of human-computer interaction. The majority of the research has concentrated on hand gesture recognition from the RGB images, however, little work has been done on recognition from videos. In addition, RGB cameras are not robust in varying lighting conditions. Motivated by this, we present the video based hand gestures recognition using the depth camera and a light weight convolutional neural network (CNN) model. We constructed a dataset and then used a light weight CNN model to detect and classify hand movements efficiently. We also examined the classification accuracy with a limited number of frames in a video gesture. We compare the depth camera’s video gesture recognition performance to that of the RGB camera. We evaluate the proposed model’s performance on edge computing devices and compare to benchmark models in terms of accuracy and inference time. The proposed model results in an accuracy of 99.48% on the RGB version of test dataset and 99.18% on the depth version of test dataset. Finally, we compare the accuracy of the proposed light weight CNN model with the state-of-the hand gesture classification models.

More details:DOI: 10.1109/JSEN.2022.3181518

Shraddha Tripathi, Om Jee Pandey, Linga Reddy Cenkeramaddi, and Rajesh M. Hegde, “Optimal Active Elements Selection in RIS-Assisted Edge Networks for Improved QoS,” has been accepted for publication in the 2022 IEEE 12th Sensor Array and Multichannel Signal Processing Workshop (SAM).

Keywords:Wireless communication,Wireless sensor networks,Energy consumption, Spectral efficiency, Computational modeling,Quality of service,Reflection

Abstract:Reflective intelligent surfaces (RIS) are emerging as a promising solution to alleviate the spectral efficiency challenges and energy consumption issues of the edge networks. RIS is com-prised of programmable, passive, and low-cost electromagnetic elements that assist blockage reduction during signal propagation over wireless channels. The quality-of-service (QoS) challenges over RIS-assisted edge networks can further be improved by the efficient utilization of RIS elements. In particular, in this work, the impacts of RIS elements’ amplitude and practical phase-shift model are investigated on the overall network sum rate and required transmit power. Moreover, considering the practical reflection model of RIS, computation of the optimal number of active RIS elements is also performed by putting constraints over QoS conditions. Extensive experimental results demonstrate a significant performance improvement using the proposed method when compared to ideal phase shift and random active elements selection models.

More details:DOI: 10.1109/SAM53842.2022.9827788

Nikethan Reddy Beeram, Srinivas Boppu, and Linga Reddy Cenkeramaddi, “Activity Classification of an Unmanned Aerial Vehicle Using Tsetlin Machine,” has been accepted for publication in the First IEEE International Symposium on Tsetlin Machine, 2022.

Keywords:Support vector machines, Privacy, Surveillance, Radar, Autonomous aerial vehicles, Security,Task analysis

Abstract:The activity classification for aerial vehicles plays a vital role in privacy monitoring and security surveillance applications, which is crucial and valuable in modern times. This paper presents the Tsetlin Machine model for aerial vehicle’s activity classification using the mm-Wave frequency modulated continuous wave (FMCW) Radar data. The proposed Tsetlin Machine (TM) model is based on propositional logic, which is much more transparent and lighter than the existing models. It can also be easily transferred to hardware, making it more useful even in practical circumstances. Furthermore, the model has a 92.5% accuracy in activity classification, which is close to other lightweight classification models like logistic regression, light gradient boosting machine (GBM) and support vector machine (SVM). Furthermore, the proposed model’s accuracy is much better than the pre-trained models such as VGG16, ResNet50, and InceptionResNet with at least 98× reduction in memory size.

More details:DOI: 10.1109/ISTM54910.2022.00022

Samala Jagadheesh, P. Veda Bhanu, Soumya J, and Lina Reddy.
Cenkeramaddi “Reinforcement Learning based Fault-Tolerant Routing Algorithm for Mesh based NoC and its FPGA Implementation” has been accepted for publication in IEEE Access (2022).

Keywords:Routing,Topology,Fault tolerant systems,Fault tolerance,Heuristic algorithms, Machine learning algorithms,Field programmable gate arrays

Abstract:Network-on-Chip (NoC) has emerged as the most promising on-chip interconnection framework in Multi-Processor System-on-Chips (MPSoCs) due to its efficiency and scalability. In the deep sub-micron level, NoCs are vulnerable to faults, which leads to the failure of network components such as links and routers. Failures in NoC components diminish system efficiency and reliability. This paper proposes a Reinforcement Learning based Fault-Tolerant Routing (RL-FTR) algorithm to tackle the routing issues caused by link and router faults in the mesh-based NoC architecture. The efficiency of the proposed RL-FTR algorithm is examined using System-C based cycle-accurate NoC simulator. Simulations are carried out by increasing the number of links and router faults in various sizes of mesh. Followed by simulations, real-time functioning of the proposed RL-FTR algorithm is observed using the FPGA implementation. Results of the simulation and hardware shows that the proposed RL-FTR algorithm provides an optimal routing path from the source router to the destination router.

More details:DOI: 10.1109/ACCESS.2022.3168992

B. Pardhasaradhi and L. R. Cenkeramaddi “GPS Spoofing Detection and Mitigation for Drones using Distributed Radar Tracking and Fusion,” has been accepted for publication in IEEE Sensors Journal (2022).

Keywords:Radar tracking,Target tracking,Global Positioning System, Sensors, Drones, Radar, Satellites

Abstract:In today’s world, Global positioning system (GPS)-based navigation is inexpensive for providing position, velocity, and time (PVT) information. GPS receivers are widely used on unmanned aerial vehicles (UAVs), and these targets are vulnerable to deliberate interference such as spoofing. In this paper, GPS spoofing detection and mitigation for UAVs are proposed using distributed radar ground stations equipped with a local tracker. In the proposed approach, UAVs and local trackers are linked to the fusion node. The UAVs estimate their position and covariance using the extended Kalman filter framework and send it to a fusion node as primary data. Simultaneously, the time-varying kinematics of the UAVs are estimated using the extended Kalman filter and global nearest neighbor association tracker frameworks, and this data is transmitted to the central fusion node as secondary data. A track-to-track association is proposed to detect spoofing attacks using available primary and secondary data. After detecting the spoofing attack, the secondary data is subjected to a correlation-free fusion. We propose using this fused state as a control input to the UAVs to mitigate the spoofing attack. The spoofing scenario results show that using the predicted fusion state provides the same accuracy as a GPS receiver in a clean environment. Furthermore, because the innovation is calculated using the predicted fused state, there is no effect on the number of satellite signals on PRMSE. Additionally, in terms of PRMSE, radars with low measurement noise outperform radars with high measurement noise. The proposed algorithm is best suited for use in drone swarm applications.

More details:DOI: 10.1109/JSEN.2022.3168940

Aveen Dayal, Sreenivasa Reddy Yeduri, Balu Harshavardan Koduru, Rahul Kumar Jaiswal, Soumya J, Srinivas M.B., Om Jee Pandey and Linga Reddy Cenkeramaddi, “Light Weight Deep Convolutional Neural Network for Background Sound Classification in Speech Signals,” has been accepted for publication in The Journal of the Acoustical Society of America (JASA), 2022.

ABSTRACT:Recognizing background information in human speech signals is a task that is extremely useful in a wide range of practical applications, and many articles on background sound classification have been published. It has not, however, been addressed with background embedded in real-world human speech signals. Thus, this work proposes a lightweight deep convolutional neural network (CNN) in conjunction with spectrograms for an efficient background sound classification with practical human speech signals. The proposed model classifies 11 different background sounds such as airplane, airport, babble, car, drone, exhibition, helicopter, restaurant, station, street, and train sounds embedded in human speech signals. The proposed deep CNN model consists of four convolution layers, four max-pooling layers, and one fully connected layer. The model is tested on human speech signals with varying signal-to-noise ratios (SNRs). Based on the results, the proposed deep CNN model utilizing spectrograms achieves an overall background sound classification accuracy of 95.2% using the human speech signals with a wide range of SNRs. It is also observed that the proposed model outperforms the benchmark models in terms of both accuracy and inference time when evaluated on edge computing devices.

More details: https://doi.org/10.1121/10.0010257