Han Zhang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Han Zhang has a total of 59 co-authors. Here are the top co-authors by number of shared publications:
- Chuan Huang
Shenzhen Institute for Advanced Study, University of Electronic Science and Technology of China, Shenzhen, China
7 shared publications - Shuguang Cui
The Chinese University of Hong Kong
6 shared publications - Jianhao Huang
The University of Hong Kong
4 shared publications - Shiyin Li
China University of Mining and Technology
4 shared publications - Shuai Ma
Peng Cheng Laboratory, Shenzhen, China
4 shared publications - Yue Gao
Fudan University
4 shared publications
Latest Publications
- 7 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Wireless Communications
- Vol. 25  (15361276) , 14766-14780
Adaptive Source–Channel Coding for Multi-User Semantic and Data Communications
This paper considers a multi-user semantic and data communication (MU-SemDaCom) system, where a base station (BS) simultaneously serves users with different semantic and data tasks through a downlink multi-user multiple-input single-output (MU-MISO) channel. The coexistence of heterogeneous communication tasks, diverse channel conditions, and the requirements for digital compatibility poses significant challenges to the efficient design of MU-SemDaCom systems. To address these issues, we propose a multi-user adaptive source-channel coding (MU-ASCC) framework that adaptively optimizes deep neural network (DNN)-based source coding, digital channel coding, and superposition broadcasting according to the channel conditions. First, we employ a data-regression method to approximate the end-to-end (E2E) semantic and data distortions, for which no closed-form expressions exist due to the complex coupling between DNN-based source coding and channel codes. The obtained logistic formulas decompose the E2E distortion as the addition of the source and channel distortion terms, in which the logistic parameter variations are task-dependent and jointly determined by both the DNN and channel parameters. Then, based on the derived formulas, we formulate a weighted-sum E2E distortion minimization problem that jointly optimizes the source-channel coding rates, power allocation, and beamforming vectors for both the data and semantic users. Finally, an alternating optimization (AO) framework is developed, where the adaptive rate optimization is solved using the subgradient descent method, while the joint power and beamforming is addressed via the uplink-downlink duality (UDD) technique. Simulation results demonstrate that, compared with the conventional separate source-channel coding (SSCC) and deep joint source-channel coding (DJSCC) schemes that are designed for a single task, the proposed MU-ASCC scheme achieves simultaneous improvements in both the data recovery and semantic task performance.
- 7 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Wireless Communications
- Vol. 25  (15361276) , 14766-14780
Adaptive Source–Channel Coding for Multi-User Semantic and Data Communications
This paper considers a multi-user semantic and data communication (MU-SemDaCom) system, where a base station (BS) simultaneously serves users with different semantic and data tasks through a downlink multi-user multiple-input single-output (MU-MISO) channel. The coexistence of heterogeneous communication tasks, diverse channel conditions, and the requirements for digital compatibility poses significant challenges to the efficient design of MU-SemDaCom systems. To address these issues, we propose a multi-user adaptive source-channel coding (MU-ASCC) framework that adaptively optimizes deep neural network (DNN)-based source coding, digital channel coding, and superposition broadcasting according to the channel conditions. First, we employ a data-regression method to approximate the end-to-end (E2E) semantic and data distortions, for which no closed-form expressions exist due to the complex coupling between DNN-based source coding and channel codes. The obtained logistic formulas decompose the E2E distortion as the addition of the source and channel distortion terms, in which the logistic parameter variations are task-dependent and jointly determined by both the DNN and channel parameters. Then, based on the derived formulas, we formulate a weighted-sum E2E distortion minimization problem that jointly optimizes the source-channel coding rates, power allocation, and beamforming vectors for both the data and semantic users. Finally, an alternating optimization (AO) framework is developed, where the adaptive rate optimization is solved using the subgradient descent method, while the joint power and beamforming is addressed via the uplink-downlink duality (UDD) technique. Simulation results demonstrate that, compared with the conventional separate source-channel coding (SSCC) and deep joint source-channel coding (DJSCC) schemes that are designed for a single task, the proposed MU-ASCC scheme achieves simultaneous improvements in both the data recovery and semantic task performance.
- 31 January 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Letters
- Vol. 8  (3) , 1-4
Compressed Sensing-Based Data Acquisition via Intelligent Vehicles
In this letter, we propose a compressed sensing (CS)-based data acquisition method for intelligent vehicles. We consider a wide-area monitoring scenario, where an intelligent vehicle is used to acquire data of the area during a period of time. We formulate an optimization problem to minimize the data recovery error and the travel distance of the intelligent vehicle and develop an efficient algorithm to find an appropriate trajectory. The experimental results demonstrate the feasibility and improvement of the proposed method compared to the existing methods.
- 31 January 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Letters
- Vol. 8  (3) , 1-4
Compressed Sensing-Based Data Acquisition via Intelligent Vehicles
In this letter, we propose a compressed sensing (CS)-based data acquisition method for intelligent vehicles. We consider a wide-area monitoring scenario, where an intelligent vehicle is used to acquire data of the area during a period of time. We formulate an optimization problem to minimize the data recovery error and the travel distance of the intelligent vehicle and develop an efficient algorithm to find an appropriate trajectory. The experimental results demonstrate the feasibility and improvement of the proposed method compared to the existing methods.
- 22 September 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 73  (2) , 2924-2929
Fast Compressed Wideband Spectrum Sensing
Compressed wideband spectrum sensing has attracted much interest in recent years, which enables flexible spectrum sharing to improve the efficiency of scarce frequency resource. Despite the great potential for sub-Nyquist-rate sampling, existing high-accurate compression sensing (CS) methods unfortunately incur the extremely high computational complexity, e.g., in recovering the sparse signal or estimating the a priori information on sparsity. This creates a serious challenge in deploying real-time wideband sensing in the resource constraint platforms. In this work, we develop a fast compressed spectrum sensing method, which achieves accurate performance but also greatly reduces the computational complexity. Our new method jointly exploits the low-rank and sparse properties of a sub-Nyquist measurement matrix. We first design a low-complexity sparsity estimator, by approximating a large covariance matrix with multiple small matrices. To recover the sparse spectrum, we then formulate one low-dimensional non-convex optimization problem via random orthogonal projection, which makes the CS method more computationally efficient. As demonstrated on real datasets, our method reduces the computational complexity of wideband spectrum sensing by $\sim 10\times$ ; moreover, it achieves highly accurate results without compromising the reconstruction/sensing performance. Thus, it has great promise for real-time sub-Nyquist sensing on low-complexity platforms.
- 22 September 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 73  (2) , 2924-2929
Fast Compressed Wideband Spectrum Sensing
Compressed wideband spectrum sensing has attracted much interest in recent years, which enables flexible spectrum sharing to improve the efficiency of scarce frequency resource. Despite the great potential for sub-Nyquist-rate sampling, existing high-accurate compression sensing (CS) methods unfortunately incur the extremely high computational complexity, e.g., in recovering the sparse signal or estimating the a priori information on sparsity. This creates a serious challenge in deploying real-time wideband sensing in the resource constraint platforms. In this work, we develop a fast compressed spectrum sensing method, which achieves accurate performance but also greatly reduces the computational complexity. Our new method jointly exploits the low-rank and sparse properties of a sub-Nyquist measurement matrix. We first design a low-complexity sparsity estimator, by approximating a large covariance matrix with multiple small matrices. To recover the sparse spectrum, we then formulate one low-dimensional non-convex optimization problem via random orthogonal projection, which makes the CS method more computationally efficient. As demonstrated on real datasets, our method reduces the computational complexity of wideband spectrum sensing by $\sim 10\times$ ; moreover, it achieves highly accurate results without compromising the reconstruction/sensing performance. Thus, it has great promise for real-time sub-Nyquist sensing on low-complexity platforms.
- 2 February 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cognitive Communications and Networking
- Vol. 9  (3) , 618-628
Adversarial Autoencoder Empowered Joint Anomaly Detection and Signal Reconstruction From Sub-Nyquist Samples
Anomaly detection is an essential part of spectrum monitoring applications. Malicious users and malfunctioning nodes could be identified via anomaly detection methods. Meanwhile, the spectrum bands that would be utilized in future 6G or satellite communication system settings are going to be wider than ever. Acquiring Nyquist sampled data from such a spectrum would require components with a very high sampling rate. To monitor a wide spectrum, a compressive sensing recovery algorithm combined with a sub-sampling approach could accomplish the task with a lower hardware cost. To solve the anomaly detection problem using a sub-sampled data stream, a joint signal recovery and anomaly detection solution utilizing an adversarial autoencoder (AAE) structure are proposed in this article. An AAE is constructed via an autoencoder and a discriminator weaved together. The discriminator would guide the autoencoder to drive its extracted feature onto a designed feature space, while the autoencoder would provide a reconstruction of the original Nyquist sampled signal. The proposed AAE structure could learn the distribution of the signal from either labelled or unlabelled training data, enabling it to work on both supervised and unsupervised data sets. The proposed AAE has shown superior reconstruction and detection performance on very sparse sampling scenarios.
- 2 February 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cognitive Communications and Networking
- Vol. 9  (3) , 618-628
Adversarial Autoencoder Empowered Joint Anomaly Detection and Signal Reconstruction From Sub-Nyquist Samples
Anomaly detection is an essential part of spectrum monitoring applications. Malicious users and malfunctioning nodes could be identified via anomaly detection methods. Meanwhile, the spectrum bands that would be utilized in future 6G or satellite communication system settings are going to be wider than ever. Acquiring Nyquist sampled data from such a spectrum would require components with a very high sampling rate. To monitor a wide spectrum, a compressive sensing recovery algorithm combined with a sub-sampling approach could accomplish the task with a lower hardware cost. To solve the anomaly detection problem using a sub-sampled data stream, a joint signal recovery and anomaly detection solution utilizing an adversarial autoencoder (AAE) structure are proposed in this article. An AAE is constructed via an autoencoder and a discriminator weaved together. The discriminator would guide the autoencoder to drive its extracted feature onto a designed feature space, while the autoencoder would provide a reconstruction of the original Nyquist sampled signal. The proposed AAE structure could learn the distribution of the signal from either labelled or unlabelled training data, enabling it to work on both supervised and unsupervised data sets. The proposed AAE has shown superior reconstruction and detection performance on very sparse sampling scenarios.
- 16 August 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 10  (1) , 42-56
Dynamic Clustering and Resource Allocation Using Deep Reinforcement Learning for Smart-Duplex Networks
Ultra dense networks (UDNs) with smart-duplex (SD), which allows the base stations (BSs) to flexibly switch between the half-duplex (HD) and full-duplex (FD), are expected to support high-density transmissions. However, to centrally handle a large network is costly, while distributed processing may suffer from the severe performance loss due to the complicated intercell interferences in the UDNs. This article aims to balance the system performance and clustering cost of the SD UDNs by dividing all small cells into several clusters. A Markov decision process (MDP) problem is formulated to maximize the average weighted sum of network throughput and clustering cost for all clusters. To approximately solve this problem, we first adopt an affinity propagation method to determine the number of clusters and the center of each cluster. Then, by treating small cells as agents, the original MDP problem is proved to be equivalent to a multiagent MDP to maximize the average reward of all small cells. Next, a multiagent deep reinforcement learning (DRL) is proposed to jointly implement the dynamic clustering for noncenter small cells, resource allocation, and duplex mode selection. Simulation results show that SD has prominent advantages over both the HD and FD in UDNs, and the proposed multiagent DRL outperforms other clustering schemes under the considered scenarios.
- 16 August 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 10  (1) , 42-56
Dynamic Clustering and Resource Allocation Using Deep Reinforcement Learning for Smart-Duplex Networks
Ultra dense networks (UDNs) with smart-duplex (SD), which allows the base stations (BSs) to flexibly switch between the half-duplex (HD) and full-duplex (FD), are expected to support high-density transmissions. However, to centrally handle a large network is costly, while distributed processing may suffer from the severe performance loss due to the complicated intercell interferences in the UDNs. This article aims to balance the system performance and clustering cost of the SD UDNs by dividing all small cells into several clusters. A Markov decision process (MDP) problem is formulated to maximize the average weighted sum of network throughput and clustering cost for all clusters. To approximately solve this problem, we first adopt an affinity propagation method to determine the number of clusters and the center of each cluster. Then, by treating small cells as agents, the original MDP problem is proved to be equivalent to a multiagent MDP to maximize the average reward of all small cells. Next, a multiagent deep reinforcement learning (DRL) is proposed to jointly implement the dynamic clustering for noncenter small cells, resource allocation, and duplex mode selection. Simulation results show that SD has prominent advantages over both the HD and FD in UDNs, and the proposed multiagent DRL outperforms other clustering schemes under the considered scenarios.
- Open Access
- 11 August 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Signal Processing
- Vol. 70  (1053587X) , 4225-4238
Approaching Sub-Nyquist Boundary: Optimized Compressed Spectrum Sensing Based on Multicoset Sampler for Multiband Signal
Compressed spectrum sensing naturally pursues the use of fewer sampling resources to achieve spectrum support reconstruction and signal recovery. The theoretical lower boundary of averaging sampling rate to recover the multiband signal has been proved to be twice the Landau rate. However, it is still unreachable in practice. Based on the multicoset sampling architecture, this paper analyzes the influencing factors on perfect spectrum reconstruction from three aspects: data model, sampling pattern and greedy reconstruction algorithms, for which practical and feasible optimization schemes are proposed. To reduce redundant reconstruction for the multiple measurement vectors (MMV) signal model, a block MMV model is proposed to improve the accuracy in the spectrum support set reconstruction process. A sampling pattern selection algorithm is proposed to ensure a higher success rate of the spectrum support reconstruction to optimize the sensing matrix. We also deduced the representation of the signal and noise energy in the reconstructed spectrum based on the mathematical model of compressed sensing. We thus proposed a non-orthogonal double-threshold matching pursuit algorithm to avoid a high false-alarm rate due to the manually set converging conditions for matching pursuit algorithms. Numerical experiments on real-world wideband signals are carried out to demonstrate the feasibility and advantages of the proposed approaches. With integrated optimization, the required sampling density to ensure perfect reconstruction is approaching the sub-Nyquist sampling boundary.
- 11 August 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Signal Processing
- Vol. 70  (1053587X) , 4225-4238
Open AccessApproaching Sub-Nyquist Boundary: Optimized Compressed Spectrum Sensing Based on Multicoset Sampler for Multiband Signal
Compressed spectrum sensing naturally pursues the use of fewer sampling resources to achieve spectrum support reconstruction and signal recovery. The theoretical lower boundary of averaging sampling rate to recover the multiband signal has been proved to be twice the Landau rate. However, it is still unreachable in practice. Based on the multicoset sampling architecture, this paper analyzes the influencing factors on perfect spectrum reconstruction from three aspects: data model, sampling pattern and greedy reconstruction algorithms, for which practical and feasible optimization schemes are proposed. To reduce redundant reconstruction for the multiple measurement vectors (MMV) signal model, a block MMV model is proposed to improve the accuracy in the spectrum support set reconstruction process. A sampling pattern selection algorithm is proposed to ensure a higher success rate of the spectrum support reconstruction to optimize the sensing matrix. We also deduced the representation of the signal and noise energy in the reconstructed spectrum based on the mathematical model of compressed sensing. We thus proposed a non-orthogonal double-threshold matching pursuit algorithm to avoid a high false-alarm rate due to the manually set converging conditions for matching pursuit algorithms. Numerical experiments on real-world wideband signals are carried out to demonstrate the feasibility and advantages of the proposed approaches. With integrated optimization, the required sampling density to ensure perfect reconstruction is approaching the sub-Nyquist sampling boundary.
- 12 April 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Wireless Communications
- Vol. 21  (10) , 8205-8215
Machine Learning Empowered Spectrum Sensing Under a Sub-Sampling Framework
Compressive sensing (CS) is a technique frequently adopted in wireless communications. By utilizing CS, a receiver could sense the state of channels with sub-Nyquist analog to digital converters when signals are sparse. Traditional CS methods struggle with non-sparse signals due to their intrinsic sparsity assumption. Therefore, we propose using deep learning (DL) to solve the vector support recovery problem with channels’ high occupancy. The simulation results show that the proposed CS framework powered by DL can perform better than a traditional CS analytical benchmark, both in high and low channel occupation regions. We also observe that the ML can work under a lower sampling rate than traditional CS methods. To process data sampled with high channel numbers, a divide and conquer tactic is implemented.
- 12 April 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Wireless Communications
- Vol. 21  (10) , 8205-8215
Machine Learning Empowered Spectrum Sensing Under a Sub-Sampling Framework
Compressive sensing (CS) is a technique frequently adopted in wireless communications. By utilizing CS, a receiver could sense the state of channels with sub-Nyquist analog to digital converters when signals are sparse. Traditional CS methods struggle with non-sparse signals due to their intrinsic sparsity assumption. Therefore, we propose using deep learning (DL) to solve the vector support recovery problem with channels’ high occupancy. The simulation results show that the proposed CS framework powered by DL can perform better than a traditional CS analytical benchmark, both in high and low channel occupation regions. We also observe that the ML can work under a lower sampling rate than traditional CS methods. To process data sampled with high channel numbers, a divide and conquer tactic is implemented.
- 29 March 2022
- Published by SAE International in SAE International Journal of Advances and Current Practices in Mobility
- Vol. 1
Comprehensive Electric Motor Cooling Modeling
A comprehensive 3D Computational Fluid Dynamics (CFD) with conjugate heat transfer (CHT) tool was developed in-house for a Tesla Model 3 electric motor. To accurately predict the power loss (heat generation) inside the electric motor, the electromagnetic process was solved to obtain the spatial-dependent power loss in the rotor, stator, and windings. CFD was utilized for simulating the coolant oil flow using the multiphase Volume of Fluid (VOF) approach and Finite Element Analysis (FEA) was used for simulating the thermal process within the solid domains. These three separate analysis modules (electromagnetic, fluid flow and thermal solid) were coupled strongly to enable two-way interactions. Thermal results obtained from the final converged simulations were compared to the test data obtained from the thermocouple measurements for the two most representative operating points of this e-motor and showed reasonable predictions with similar trend as observed in the test. The overall variations were under 2 °C at all thermocouple locations which was within the margin of error of a standard K-type thermocouple.
- 29 March 2022
- Published by SAE International in SAE International Journal of Advances and Current Practices in Mobility
- Vol. 1
Comprehensive Electric Motor Cooling Modeling
A comprehensive 3D Computational Fluid Dynamics (CFD) with conjugate heat transfer (CHT) tool was developed in-house for a Tesla Model 3 electric motor. To accurately predict the power loss (heat generation) inside the electric motor, the electromagnetic process was solved to obtain the spatial-dependent power loss in the rotor, stator, and windings. CFD was utilized for simulating the coolant oil flow using the multiphase Volume of Fluid (VOF) approach and Finite Element Analysis (FEA) was used for simulating the thermal process within the solid domains. These three separate analysis modules (electromagnetic, fluid flow and thermal solid) were coupled strongly to enable two-way interactions. Thermal results obtained from the final converged simulations were compared to the test data obtained from the thermocouple measurements for the two most representative operating points of this e-motor and showed reasonable predictions with similar trend as observed in the test. The overall variations were under 2 °C at all thermocouple locations which was within the margin of error of a standard K-type thermocouple.
- 14 January 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal on Selected Areas in Communications
- Vol. 40  (5) , 1457-1472
Noncoherent Massive Random Access for Inhomogeneous Networks: From Message Passing to Deep Learning
Massive machine-type communications (mMTC) are expected to support a large amount of randomly deployed users for short package transmissions. Noncoherent random access provides an efficient and practical multi-access protocol for mMTC, and also poses new challenges for the receiver design. In this paper, we leverage two well-known methods, i.e., message passing and deep learning, to jointly detect the user activity and the desired data for the noncoherent mMTC. First, by exploiting the exact distribution information of the received signal, a generalized approximate message passing (GAMP)-based algorithm is proposed, which is shown to jointly detect the user activity and the desired data by two modules: inter-user interference elimination and data detection for each user. Inspired by the two-module GAMP-based algorithm, we then propose a model-driven deep learning method, which utilizes the deep neural networks (DNNs) to approximate both the two modules. The loss function for training the DNNs is derived by formulating the two-module detection as an unconstrained optimization problem. Simulation results reveal that the proposed GAMP-based algorithm outperforms the proposed deep learning method when the channel distribution is perfectly known, while it suffers from a significant performance degradation for the case with imperfect channel distribution information.
- 14 January 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal on Selected Areas in Communications
- Vol. 40  (5) , 1457-1472
Noncoherent Massive Random Access for Inhomogeneous Networks: From Message Passing to Deep Learning
Massive machine-type communications (mMTC) are expected to support a large amount of randomly deployed users for short package transmissions. Noncoherent random access provides an efficient and practical multi-access protocol for mMTC, and also poses new challenges for the receiver design. In this paper, we leverage two well-known methods, i.e., message passing and deep learning, to jointly detect the user activity and the desired data for the noncoherent mMTC. First, by exploiting the exact distribution information of the received signal, a generalized approximate message passing (GAMP)-based algorithm is proposed, which is shown to jointly detect the user activity and the desired data by two modules: inter-user interference elimination and data detection for each user. Inspired by the two-module GAMP-based algorithm, we then propose a model-driven deep learning method, which utilizes the deep neural networks (DNNs) to approximate both the two modules. The loss function for training the DNNs is derived by formulating the two-module detection as an unconstrained optimization problem. Simulation results reveal that the proposed GAMP-based algorithm outperforms the proposed deep learning method when the channel distribution is perfectly known, while it suffers from a significant performance degradation for the case with imperfect channel distribution information.
- 31 December 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Wireless Communications
- Vol. 21  (7) , 5175-5190
Compressed Random Access for Noncoherent Massive Machine-Type Communications With Energy Modulation
Massive machine-type communications (mMTC) for the Internet of Things (IoT) are expected to support a large number of devices/users for short packet transmissions with low complexity and low energy consumption. By utilizing the simple while efficient noncoherent energy-based transmission scheme, this work aims to jointly detect the user activity and the desired data for mMTC. First, by exploiting the sparse characteristics of the user activity, approximation message passing (AMP) algorithm is proposed to eliminate the multi-user interference, and a denoiser is designed to minimize the mean-squared error (MSE) of the transmitted signals. Then, maximum ${a}$ posteriori (MAP) criterion is adopted to approximately detect the user activity and the desired data. By minimizing the symbol error probability of the above two-step algorithm, the power constellation for each user is designed, and it is shown to be asymptotically optimal as the number of the receiver antennas goes to infinity. Finally, simulation results reveal that the proposed noncoherent scheme outperforms the coherent one in the low SNR regime and for short packet transmissions.
- 31 December 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Wireless Communications
- Vol. 21  (7) , 5175-5190
Compressed Random Access for Noncoherent Massive Machine-Type Communications With Energy Modulation
Massive machine-type communications (mMTC) for the Internet of Things (IoT) are expected to support a large number of devices/users for short packet transmissions with low complexity and low energy consumption. By utilizing the simple while efficient noncoherent energy-based transmission scheme, this work aims to jointly detect the user activity and the desired data for mMTC. First, by exploiting the sparse characteristics of the user activity, approximation message passing (AMP) algorithm is proposed to eliminate the multi-user interference, and a denoiser is designed to minimize the mean-squared error (MSE) of the transmitted signals. Then, maximum ${a}$ posteriori (MAP) criterion is adopted to approximately detect the user activity and the desired data. By minimizing the symbol error probability of the above two-step algorithm, the power constellation for each user is designed, and it is shown to be asymptotically optimal as the number of the receiver antennas goes to infinity. Finally, simulation results reveal that the proposed noncoherent scheme outperforms the coherent one in the low SNR regime and for short packet transmissions.