Neha Dawar
Research Subjects
Neha Dawar has a total of 15 co-authors. Here are the top co-authors by number of shared publications:
- Abhishek Sehgal
Standards and Mobility Innovation Laboratory, Samsung Research America, Plano, TX, USA
2 shared publications Boon Loong Ng
Standards and Mobility Innovation Laboratory, Samsung Research America, Plano, TX, USA
2 shared publications- Nasser Kehtarnavaz
University of Texas at Dallas, USA
2 shared publications Yuming Zhu
Samsung Research America, Plano, TX, USA
2 shared publicationsCharlie Zhang
Samsung Research America, Plano, TX, USA
1 shared publication- Hao Chen
Samsung Research America, Plano, TX, USA
1 shared publication
Latest Publications
- Open Access
- 3 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 63298-63309
Enhancing Wi-Fi 7: Traffic Flow Intelligence and Multi-Link Operation for Optimal Efficiency
As the proliferation of Wi-Fi-enabled devices continues to accelerate, there is a pressing need for efficient wireless connectivity solutions. To address this challenge, the introduction of Multi-Link Operation (MLO) in Wi-Fi 7 (IEEE 802.11be) enables clients to establish connections with access points across multiple frequency bands, thereby enhancing throughput and reducing latency at the expense of increased power consumption. To balance the benefits of MLO with its potential impact on smartphone battery life and user experience, we propose an innovative approach that integrates machine learning-based network service detection for Traffic Flow Intelligence with intelligent link adaptation for MLO. Our solution utilizes machine learning to detect latency-sensitive applications with up to 99% accuracy and optimizes link settings based on the Quality of Experience (QoE) of the user relative to the application, detected by our QoE Monitoring Classifier with 99.7% recall for bad QoE and 88% accuracy overall. Notably, our approach enables users to enjoy up to 40 minutes more of high-quality video calls compared to an always-on MLO configuration on the Samsung Galaxy S24 Ultra.
- 3 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 63298-63309
Open AccessEnhancing Wi-Fi 7: Traffic Flow Intelligence and Multi-Link Operation for Optimal Efficiency
As the proliferation of Wi-Fi-enabled devices continues to accelerate, there is a pressing need for efficient wireless connectivity solutions. To address this challenge, the introduction of Multi-Link Operation (MLO) in Wi-Fi 7 (IEEE 802.11be) enables clients to establish connections with access points across multiple frequency bands, thereby enhancing throughput and reducing latency at the expense of increased power consumption. To balance the benefits of MLO with its potential impact on smartphone battery life and user experience, we propose an innovative approach that integrates machine learning-based network service detection for Traffic Flow Intelligence with intelligent link adaptation for MLO. Our solution utilizes machine learning to detect latency-sensitive applications with up to 99% accuracy and optimizes link settings based on the Quality of Experience (QoE) of the user relative to the application, detected by our QoE Monitoring Classifier with 99.7% recall for bad QoE and 88% accuracy overall. Notably, our approach enables users to enjoy up to 40 minutes more of high-quality video calls compared to an always-on MLO configuration on the Samsung Galaxy S24 Ultra.
- Open Access
- 1 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 10412-10421
A Low-Duty-Cycle and Congestion-Robust Design of Adaptive Wi-Fi 7 Multi-Link Operation Control Over User Side
Wi-Fi 7 has been developed recently to support newapplications such as 4K videos and AR/VR, which demand higher network throughput and lower latency. One of its key features, Multi-link Operation (MLO), enables Wi-Fi 7 to utilize multiple carrier frequencies, enhancing throughput and reducing latency. Recent research has extensively covered MAC-layer architecture and multi-link channel access algorithm design for MLO implementation, but relatively few studies address the power efficiency problem when MLO is implemented in the user devices under dynamic traffic conditions. For user devices, power efficiency is critical due to stricter power budgets, and hence, a minimal radio activation time (i.e., duty cycle) is needed. We have devised an adaptive algorithm that dynamically adjusts MLO configurations based on periodic measurements at the MAC/IP layers. Our simulations using the NS-3 platform demonstrate that our approach maintains MLO’s congestion-resistant throughput advantages while significantly reducing the radio’s duty cycle compared to a static MLO mode. This adaptation makes MLO more feasible for power-sensitive user devices.
- 1 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 10412-10421
Open AccessA Low-Duty-Cycle and Congestion-Robust Design of Adaptive Wi-Fi 7 Multi-Link Operation Control Over User Side
Wi-Fi 7 has been developed recently to support newapplications such as 4K videos and AR/VR, which demand higher network throughput and lower latency. One of its key features, Multi-link Operation (MLO), enables Wi-Fi 7 to utilize multiple carrier frequencies, enhancing throughput and reducing latency. Recent research has extensively covered MAC-layer architecture and multi-link channel access algorithm design for MLO implementation, but relatively few studies address the power efficiency problem when MLO is implemented in the user devices under dynamic traffic conditions. For user devices, power efficiency is critical due to stricter power budgets, and hence, a minimal radio activation time (i.e., duty cycle) is needed. We have devised an adaptive algorithm that dynamically adjusts MLO configurations based on periodic measurements at the MAC/IP layers. Our simulations using the NS-3 platform demonstrate that our approach maintains MLO’s congestion-resistant throughput advantages while significantly reducing the radio’s duty cycle compared to a static MLO mode. This adaptation makes MLO more feasible for power-sensitive user devices.
- Open Access
- 18 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 41084-41101
WhereArtThou: A WiFi-RTT-Based Indoor Positioning System
Accurate indoor positioning is critical to a variety of use cases including tracking, proximity, mapping, and navigation. Existing positioning methods are either inaccurate (e.g. RSSI-based ranging), impractical (e.g. fingerprinting), or uncommon (e.g. Ultra-Wide Band/UWB).WiFi Round-Trip Time (RTT) marries the ubiquity of WiFi infrastructure with the accuracy of such ranging mechanisms as those used in UWB. We present WhereArtThou , a commercial-grade, plug-and-play indoor positioning solution (IPS) based on WiFi RTT. Our base algorithm uses an extended Kalman filter with a random walk motion model (EKF-RW) which relies solely on RTT distance measurements. We propose two EKF components to enhance positioning accuracy. First, as distance measurements can further be fused with inertial sensor readings, we propose a step-and-heading-based filter (EKF-SH) when such readings are available. We devise a method to fit the non-Gaussian step error with a Gaussian random variable to remain within the computationally-efficient Kalman filtering framework. Second, we define a distance-dependent measurement model to match the true statistics of the measurement noise and approach optimal position estimation. Moreover, we measure the gain from the proposed enhancements not only through the almost-exclusively-used error metric in the indoor positioning literature, the Euclidean distance and its variations, but also through metrics commonly used in satellite and maritime navigation, the cross-track and along-track errors, in addition to a set of metrics of our own definition. Finally, we show that RTT is highly susceptible to the human body holding the device measuring it, and we case-study the impact of human-body blockage on positioning and ranging errors. We test our algorithms on over 18 hours of walking data collected on different devices, in different locations, and with different users, and we observe that the EKF-RW and EKF-SH achieve 90th percentile distance errors of 1.65 m and 1.45 m, and 90th percentile cross-track errors of 0.85 m and 1.55 m, making our solution primed for commercial deployment.
- 18 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 41084-41101
Open AccessWhereArtThou: A WiFi-RTT-Based Indoor Positioning System
Accurate indoor positioning is critical to a variety of use cases including tracking, proximity, mapping, and navigation. Existing positioning methods are either inaccurate (e.g. RSSI-based ranging), impractical (e.g. fingerprinting), or uncommon (e.g. Ultra-Wide Band/UWB).WiFi Round-Trip Time (RTT) marries the ubiquity of WiFi infrastructure with the accuracy of such ranging mechanisms as those used in UWB. We present WhereArtThou , a commercial-grade, plug-and-play indoor positioning solution (IPS) based on WiFi RTT. Our base algorithm uses an extended Kalman filter with a random walk motion model (EKF-RW) which relies solely on RTT distance measurements. We propose two EKF components to enhance positioning accuracy. First, as distance measurements can further be fused with inertial sensor readings, we propose a step-and-heading-based filter (EKF-SH) when such readings are available. We devise a method to fit the non-Gaussian step error with a Gaussian random variable to remain within the computationally-efficient Kalman filtering framework. Second, we define a distance-dependent measurement model to match the true statistics of the measurement noise and approach optimal position estimation. Moreover, we measure the gain from the proposed enhancements not only through the almost-exclusively-used error metric in the indoor positioning literature, the Euclidean distance and its variations, but also through metrics commonly used in satellite and maritime navigation, the cross-track and along-track errors, in addition to a set of metrics of our own definition. Finally, we show that RTT is highly susceptible to the human body holding the device measuring it, and we case-study the impact of human-body blockage on positioning and ranging errors. We test our algorithms on over 18 hours of walking data collected on different devices, in different locations, and with different users, and we observe that the EKF-RW and EKF-SH achieve 90th percentile distance errors of 1.65 m and 1.45 m, and 90th percentile cross-track errors of 0.85 m and 1.55 m, making our solution primed for commercial deployment.
- 29 October 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Letters
- Vol. 3 (1) , 1-4
Data Augmentation in Deep Learning-Based Fusion of Depth and Inertial Sensing for Action Recognition
This article covers a deep learning-based decision fusion approach for action or gesture recognition via simultaneous utilization of a depth camera and a wearable inertial sensor. The deep learning approach involves using a convolutional neural network (CNN) for depth images captured by a depth camera and a combination of CNN and long short–term memory network for inertial signals captured by a wearable inertial sensor, followed by a decision-level fusion. Due to the limited size of the training data, a data augmentation procedure is carried out by generating depth images corresponding to different orientations of the depth camera and by generating inertial signals corresponding to different orientations of the inertial sensor placement on the body. The results obtained indicate the positive impact of the decision-level fusion as well as the data augmentation on the recognition accuracies.
- 29 October 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Letters
- Vol. 3 (1) , 1-4
Data Augmentation in Deep Learning-Based Fusion of Depth and Inertial Sensing for Action Recognition
This article covers a deep learning-based decision fusion approach for action or gesture recognition via simultaneous utilization of a depth camera and a wearable inertial sensor. The deep learning approach involves using a convolutional neural network (CNN) for depth images captured by a depth camera and a combination of CNN and long short–term memory network for inertial signals captured by a wearable inertial sensor, followed by a decision-level fusion. Due to the limited size of the training data, a data augmentation procedure is carried out by generating depth images corresponding to different orientations of the depth camera and by generating inertial signals corresponding to different orientations of the inertial sensor placement on the body. The results obtained indicate the positive impact of the decision-level fusion as well as the data augmentation on the recognition accuracies.
- 28 September 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 18 (23) , 9660-9668
Action Detection and Recognition in Continuous Action Streams by Deep Learning-Based Sensing Fusion
This paper presents a deep learning-based sensing fusion system to detect and recognize actions of interest from continuous action streams which contain actions of interest occurring continuously and randomly among arbitrary actions of non-interest. The sensors used in the fusion system consist of a depth camera and a wearable inertial sensor. A convolutional neural network is utilized for depth images obtained from the depth sensor and a combination of convolutional neural network and long short term memory network is utilized for inertial signals obtained from the inertial sensor. Each sensing modality first performs segmentation of all actions and then detection of actions of interest for a particular application. A decision-level fusion of the two sensing modalities is carried out to achieve the recognition of the detected actions of interest. The developed fusion system is examined for two applications: one involving transition movements for home healthcare monitoring and the other involving smart TV hand gestures. The results obtained show the effectiveness of the developed fusion system in dealing with realistic continuous action streams.
- 28 September 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 18 (23) , 9660-9668
Action Detection and Recognition in Continuous Action Streams by Deep Learning-Based Sensing Fusion
This paper presents a deep learning-based sensing fusion system to detect and recognize actions of interest from continuous action streams which contain actions of interest occurring continuously and randomly among arbitrary actions of non-interest. The sensors used in the fusion system consist of a depth camera and a wearable inertial sensor. A convolutional neural network is utilized for depth images obtained from the depth sensor and a combination of convolutional neural network and long short term memory network is utilized for inertial signals obtained from the inertial sensor. Each sensing modality first performs segmentation of all actions and then detection of actions of interest for a particular application. A decision-level fusion of the two sensing modalities is carried out to achieve the recognition of the detected actions of interest. The developed fusion system is examined for two applications: one involving transition movements for home healthcare monitoring and the other involving smart TV hand gestures. The results obtained show the effectiveness of the developed fusion system in dealing with realistic continuous action streams.
- Open Access
- 1 January 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 6, 7019-7028
Real-Time Continuous Detection and Recognition of Subject-Specific Smart TV Gestures via Fusion of Depth and Inertial Sensing
This paper presents a real-time detection and recognition approach to identify actions of interest involved in the smart TV application from continuous action streams via simultaneous utilization of a depth camera and a wearable inertial sensor. Continuous action streams mean when actions of interest are performed continuously and randomly among arbitrary actions of non-interest. The developed approach consists of a detection part and a recognition part. In the detection part, two support vector data descriptor classifiers corresponding to the two sensing modalities are used to separate actions of interest from actions of non-interest in continuous action streams. The actions detected as actions of interest by both of the sensing modalities are then passed to the recognition part. In this part, actions of interest are classified by fusing the decisions from two collaborative representation classifiers, one classifier using skeleton joint positions and the other classifier using inertial signals. The developed approach is applied to the hand gestures in the smart TV application. The experimental results obtained indicate the effectiveness of the developed approach to detect and recognize smart TV gestures in continuous action streams.
- 1 January 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 6, 7019-7028
Open AccessReal-Time Continuous Detection and Recognition of Subject-Specific Smart TV Gestures via Fusion of Depth and Inertial Sensing
This paper presents a real-time detection and recognition approach to identify actions of interest involved in the smart TV application from continuous action streams via simultaneous utilization of a depth camera and a wearable inertial sensor. Continuous action streams mean when actions of interest are performed continuously and randomly among arbitrary actions of non-interest. The developed approach consists of a detection part and a recognition part. In the detection part, two support vector data descriptor classifiers corresponding to the two sensing modalities are used to separate actions of interest from actions of non-interest in continuous action streams. The actions detected as actions of interest by both of the sensing modalities are then passed to the recognition part. In this part, actions of interest are classified by fusing the decisions from two collaborative representation classifiers, one classifier using skeleton joint positions and the other classifier using inertial signals. The developed approach is applied to the hand gestures in the smart TV application. The experimental results obtained indicate the effectiveness of the developed approach to detect and recognize smart TV gestures in continuous action streams.