In dyadic interactions, facial reactions are crucial for conveying an individuals' responses to their conversational partners. Individuals may exhibit varied but appropriate facial reactions (AFRs) when perceiving the same behavioral expression. Although some recent methods can already respond multiple appropriate facial reactions to the given human speaker behaviors, the AFRs generated by these methods often fail to adequately preserve crucial head motions, leading to visual jitter and unnatural transitions between generated AFR segments. In this paper, we propose a novel and generic PFLPosNet framework which addresses the aforementioned problems at both pre-processing and post-processing stages, where a new pose-aware face behavior localization method PFL is introduced to retain the head pose displacement information from the source data. In addition, the framework proposes a real-time head pose adjustment method, PosNet, to ensure continuity and smoothness in the visual output of the model when using data with correct head pose displacement. Experimental results demonstrate that our approach not only generates more coherent and natural facial reaction sequences but also significantly outperforms existing online MAFRG methods in terms of continuity and smoothness. Our code is made available at https://github.com/rainforcetime/PFLPosNet.

Unsourced multiple access (UMA) is a promising technology for the sixth generation (6G) mobile networks to fulfil the requirements of massive random access or massive small data transmission. However, the performance evaluation of UMA for non-orthogonal data transmission with real user active detection and real channel estimation is not thoroughly considered for fading channels. In this paper realistic user active detection and realistic channel estimation is carried out for unsourced sparse multiple access (USMA) scheme where sparse interleaver division multiple access is employed. The estimated channels of active users from preamble part are applied for multi-user signal detection for data transmission part. It is found that the multi-user detection performance loss of USMA compared to the ideal channel estimation is marginal in both additive white Gaussian noise (AWGN) channel and flat fading channel under the reasonable system parameter setting.

By using a large modulation index a Gaussian spectrum can be obtained by constant-envelope orthogonal frequency division multiplexing (CE-OFDM) systems, thus an excellent ambiguity function for radar systems is achieved. However, the bit error rate performance for communication will degrade significantly due to large modulation index applied. To solve this problem in this paper a phase-clipping CE-OFDM waveform is proposed for integrated sensing and communication systems. By clipping the phase at transmitter side and using phase compensation at receiver side, the bit error rate performance is enhanced greatly with slightly performance degradation for radar systems. The proposed waveform can be an excellent candidate for 6th generation integrated sensing and communication systems.

In this paper, a spreading based multi-branch non-orthogonal multiple access (NOMA) scheme is proposed where the information bits are split into several branches with independent channel encoding. A legacy quadrature phase shift keying (QPSK) modulation with different power allocation is applied for bit to symbol mapping of each branch. After superposition of constellations from different branches, user-specific spreading sequence is applied which is used for multi-user interference suppression. At receiver side minimum mean square error (MMSE) elementary signal estimator (ESE) is used for multi-user detection. Due to independent channel decoding and QPSK as constellation where two bits have the same reliability, the quality of the log likelihood ratio for the branch with high power is high whose signals will be successfully decoded first. This will facilitate the decoding of the signals of the branch with low power. Our simulation results show NOMA using multi-branch strategy is superior to its counterpart and thus it is a strong candidate as 5G transmission scheme.

In this paper, a blind multiple user detection for multi-user shared access systems (MUSA) is introduced for massive machine type communication (mMTC). Without relying on reference signal to estimate the channel state information, user traffic signal is detected and decoded via a unique blind detection algorithm. The successfully decoded signal is employed as ``pilot'' for channel estimation which would be refined as more signals are successfully decoded. High-overloading can be achieved for grant-free transmission of random signature selection, with relatively low complexity receiver, which are preferred by mMTC.

Non-orthogonal multiple access (NOMA) is one of the promising radio access techniques for further cellular enhancements toward 5G. Compared to orthogonal multiple access (OMA) such as orthogonal frequency-division multiple access (OFDMA), large performance gains were confirmed via system-level simulations. However, NOMA link-level simulations and the design of the receiver remain of great importance to validate NOMA performance gains. In this paper, we evaluate downlink NOMA link-level performance with multiple receiver designs and propose a novel NOMA transmitter and receiver design, where the signals of multi-users are jointly modulated at transmitter side and detected at receiver side. The predominant advantage of the proposed scheme is that at receiver side interference cancellation to the interference signal is not needed, thus low complexity is achieved. The performances of codeword-level SIC, symbol-level SIC and the proposed receiver are evaluated and compared with ideal SIC. Simulation results show that compared with ideal SIC, downlink NOMA link-level performance depends on actual receiver design and the difference in the power ratio split between the cell edge user and cell center user. In particular, it is shown that codeword-level SIC and the proposed receiver can both provide a good performance even when the power ratio difference between the cell center user and cell edge user is small and with real channel estimation.

In relay systems, the quality of the received signal at a mobile user is quite different from that at a relay node making it a good application scenario for hierarchical modulation. In conventional multi-user transmission schemes for relay systems where hierarchical modulation is applied, high level power is always assigned to the same mobile user, at both the source and relay node. The performance of one mobile user is always superior to that of the other; therefore, the performance of mobile users is unbalanced. In this paper, we propose a novel multiuser transmission scheme that implements different hierarchical modulation signal in a relay node to achieve balanced bit error rate performance among the mobile users. The strategy we employ is that when high power is allocated to a mobile user at a base station, low power is allocated to the hierarchical modulation signal of the same mobile user at the relay node. Hence, the strengths of the signals at different mobile users, which are a combination of the signals from the base station and relay node, are comparable. Then mobile users can achieve excellent performance simultaneously, and a balanced performance can be obtained by using proposed scheme. Simulation results show effectiveness of proposed scheme.

This paper investigates the optimization of precoders for uplink multi-user wireless communication systems, where the transmitters and receiver are equipped with multiple antennas. It is assumed that the centralized receiver has perfect knowledge of each uplink channel and designs the precoders for all users. The precoders are quantized to the vectors of a codebook, and the precoder vector index (PVI) is sent to each user through a limited channel. Since interference exists among users, the traditional single-user optimization-based precoding scheme exhibits poor performance. To improve the multi-user uplink sum-rate performance, a scheme incorporating an iterative multiple user PVI selection algorithm is proposed that iteratively projects one user channel into the null space of the precoded channels of other users, and obtains the PVI from the projected space. Simulation results show significant performance improvement compared to the conventional scheme. We also show that the proposed scheme achieves performance near to that for an exhaustive search, and outperforms the scheme that directly quantizes the precoders calculated from un-quantized joint optimization algorithms.

This paper studies the constant amplitude codebook based feedback scheme for Tomlinson-Harashima Precoding (THP) in multi-user multiple-input-multiple-output (MU-MIMO) systems. Conventional THP precoding scheme uses non-constant amplitude codebook to encapsulate channel direction information as well as channel amplitude information. In this paper we first prove that channel amplitude information is not necessary in the user channel quantization for THP precoding with limited feedback, thus in the proposed scheme we can apply constant amplitude codebook for channel quantization. Because we do not waste feedback bits to represent the channel amplitude information in the codebook applied, the proposed scheme achieves better quantization accuracy leading to better sum-rate performance compared with the conventional one. Our analysis and the validity of the proposed scheme are verified by numerical simulations.

Coherent space-time codes perform not so excellent in high moving speed case due to the inaccurate channel estimation. However, differential space-time code is appealing in this case because no channel estimation is needed. While in low and middle signal-to-noise ratio regions differential space-time codes are inferior to the coherent ones. To achieve an excellent performance in a various moving speed environments, we develop an adaptive transmission scheme in this paper where coherent space-time code is applied in low and middle moving speed cases, while differential space-time code is used in high moving speed case. Because the proposed scheme adopts the space-time codes wisely according to the moving speed, it outperforms the scheme which purely utilizes coherent space-time code or differential space-time code and thus is suitable for various moving speed environments. Moreover, two differential space-time codes based on differential Alamouti code are proposed in this paper with the merits of low decoding complexity, suitable for general number of transmit antennas and no code rate loss. The effectiveness of the proposed scheme and the proposed differential space-time codes is supported by numerical simulations.

Recently there is a trend to use Tomlinson-Harashima preceding (THP) to achieve better performance with full channel state information known at transmit side. In general, performance of THP precoding is governed by the the minimum absolute diagonal elements of the triangle matrix which is derived from QR decomposition to the channel matrix. In this paper we propose a new THP precoding scheme in which codebook based precoding is applied to optimize the effective channel after THP precoding with the aim of maximizing the minimum absolute diagonal elements of the triangle matrix. As a result, about 4 to 8 dB gains are obtained by employing codebooks of different sizes. The merits of the proposed scheme are low complexity, flexible tradeoff between complexity and performance by employing codebook with different sizes, and friendly to hardware implementation because of its fixed complexity. The theory analysis on the BER performance of THP precoding scheme is provided in this paper which is close to the simulation results. Numerical simulation results verify our conclusions.

Detection is one of the key technologies in multiple-input multiple-output system. In this paper we propose a new tree searching scheme which is flexible in the tradeoff between performance and complexity. Sorted QR decomposition based on minimum mean square error (MMSE) criteria is applied as a preprocessing to make the channel easier for tree searching and Fano bias, which is adjustable to achieve a good tradeoff between performance and complexity, is applied in the radius setting for different layer to more efficiently prune the leaves. Much smaller complexity is needed by the proposed scheme with negligible performance degradation compared with that of sphere decoding scheme, especially in the scenarios of larger number of transmitter antennas and high order modulation. Simulations accommodate our conclusions.

In multiuser MIMO-OFDM closed-loop system downlink, a proper scheduling algorithm can be used to extract the multi-user diversity by pouring more space-time-frequency (STF) resources to the user with better channel quality. For these systems, we aim to achieve further improvement without increasing the feedback overhead, and without sacrificing the performance gain the multiuser scheduling can provide. In this paper, a new space-frequency precoding scheme is proposed which makes better use of the additional diversity among different STF resources. In this scheme, the symbol streams at different STF resource are rotated before transmitting. These symbol streams should belong to the same user and have the same modulation and coding (MC) level. The rotation balances the signal to interference plus noise ratio (SINR) of each symbol stream, the average bit error rate (BER) performance is improved. With linear receiver, we proposed to use FFT basis as the rotation. Two cases occurred in space division multiple assess (SDMA) system are discussed. By the simulations in 3GPP long term evaluation (LTE) environment, the proposed scheme is proved to achieve lower BER and higher throughput than the existing systems.

Some flexible space-time codes such as thread algebraic space-time (TAST) can achieve the optimal multiplexing-diversity tradeoff. And TAST is very flexibility in adaptively adjusting its code rate. By exploiting this characteristic, we propose a novel link-adaptation scheme called AM-STC (adaptive modulation and space-time coding) scheme, which is an integration of adaptive modulation and flexible STC based on TAST code. By combining two schemes, we can get wider link adaptation against dynamic channel variation. Compared with conventional method where only adaptive modulation is used, it is clarified that our scheme has more flexibility and higher spectrum efficiency under the same BER environment

In this paper, we propose a novel method for orthogonal frequency-division multiplexing (OFDM) time and frequency synchronization by using PN sequence and barker code, time synchronization performance is improved by barker code, and we also propose a simple but effective method by setting the threshold during time synchronization. Improved frequency synchronization algorithm is presented in this paper too. The idea of multipath utilizing is the core of the frequency synchronization algorithm. The performance of the synchronization proposed here is better than that of conventional methods. Estimation precision is boosted from 10/sup -3/ to 10/sup -4/ for moderate SNR. The requirements of time and frequency synchronization for beyond 3G system based on OFDM can be well satisfied while using the proposed method.