This paper proposes Flex-QUIC, an AI-empowered quick UDP Internet connections (QUIC) enhancement framework that addresses the challenge of degraded transmission efficiency caused by the static parameterization of acknowledgment (ACK) mechanisms, loss detection, and forward error correction (FEC) in dynamic wireless networks. Unlike the standard QUIC protocol, Flex-QUIC systematically integrates machine learning across three critical modules to achieve high-efficiency operation. First, a contextual multi-armed bandit-based ACK adaptation mechanism optimizes the ACK ratio to reduce wireless channel contention. Second, the adaptive loss detection module utilizes a long short-term memory (LSTM) model to predict the reordering displacement for optimizing the packet reordering tolerance. Third, the FEC transmission scheme jointly adjusts the redundancy level based on the LSTM-predicted loss rate and congestion window state. Extensive evaluations across Wi-Fi, 5G, and satellite network scenarios demonstrate that Flex-QUIC significantly improves throughput and latency reduction compared to the standard QUIC and other enhanced QUIC variants, highlighting its adaptability to diverse and dynamic network conditions. Finally, we further discuss open issues in deploying AI-native transport protocols.

Synthetic aperture radar (SAR) radio frequency identification (RFID) localization is widely used for automated guided vehicles (AGVs) in the industrial internet of things (IIoT). However, the AGV’s speeds are limited by the phase difference (PD) of two neighboring readers. In this paper, an inertial navigation system (INS) based SAR RFID localization method (ISRL) where AGV moves nonlinearly. To relax the speed limitation, a new phase-unwrapping method based on the similarity of PDs (PU-SPD) is proposed to deal with the PD ambiguity when the AGV speed exceeds 60km/h. In localization, the gauss-newton algorithm (GN) is employed and an initial value estimation scheme based on variable substitution (IVE-VS) is proposed to improve its positioning accuracy and the convergence rate. Thus, ISRL is a combination of IVE-VS and GN. Moreover, the Cramer-Rao lower bound (CRLB) and the speed limitation is derived. Simulation results show that the ISRL can converge after two iterations, and the positioning accuracy can achieve 7.50cm at a phase noise level $\sigma=0.18$, which is 35% better than the Hyperbolic unbiased estimation localization (HyUnb).

Time synchronization is a prerequisite for ensuring determinism in time-sensitive networking (TSN). While time synchronization errors cannot be overlooked, pursuing minimal time errors may incur unnecessary costs. Using complex network theory, this study proposes a hierarchy for TSN and introduces the concept of bounded time error. A coupling model between traffic scheduling and time synchronization is established, deriving functional relationships among end-to-end delay, delay jitter, gate window, and time error. These relationships illustrate that time errors can trigger jumps in delay and delay jitter. To evaluate different time errors impact on traffic scheduling performance, an end-to-end transmission experiment scheme is designed, along with the construction of a TSN test platform implementing two representative cases. Case A is a closed TSN domain scenario with pure TSN switches emulating closed factory floor network. Case B depicts remote factory interconnection where TSN domains link via non-TSN domains composed of OpenFlow switches. Results from Case A show that delay and delay jitter on a single node are most significantly affected by time errors, up to one gating cycle. End-to-end delay jitter tends to increase with the number of hops. When the ratio of time error bound to window exceeds 10%, the number of schedulable traffic flows decreases rapidly. Case B reveals that when time error is below 1 $\mu$ s, the number of schedulable traffic flows begins to increase significantly, approaching full schedulability at errors below 0.6 $\mu$ s.

The WSN (wireless sensor network) node optimization problem faces the challenge of efficient deployment and adaptation under limited resources and a dynamically changing environment. The complex and changing deployment environment puts higher requirements on the search space, computational cost, and optimization efficiency of the algorithms. For this reason, a slime mould algorithm called SCA-SMA is proposed to solve the above problem. In SCA-SMA, a reverse Sobol sequence is used to initialize the population to increase the population diversity and improve the probability of approaching the optimal solution. To better balance local exploitation and global exploration, a dynamic selection of sine cosine update mechanism is proposed: using an optimal position selection mechanism in the global exploration phase to avoid local optima, and integrating the sine cosine algorithm in the local exploitation phase to improve the mucilage position update method, enrich the optimization search process and enhance the development capability of the algorithm. Finally, an adaptive mutation strategy can be proposed to increase the search range of the algorithm and motivate SCA-SMA to explore more promising regions. To evaluate the performance of the algorithm, SCA-SMA is experimentally validated in five different aspects. The results show that SCA-SMA is significantly competitive compared to advanced MAs. In particular, in facing the WSN node coverage problem, SCA-SMA has more obvious advantages in both average coverage and optimal coverage, which makes it possible to fully utilize the sensing range of each sensor node, while avoiding the waste of resources and the generation of monitoring blind zones.

The space-air-ground integrated network (SAGIN) is an important sixth-generation (6G) scenario that is enabled by dynamic spot beam forming by a phased array antenna (PAA). The extremely high mobility of satellites and more complicated radio resource control (RRC) have brought introduced a new challenge, and the issue of determining appropriate moments and procedures for executing handover (HO) for all users in a coverage area is urgent. The existing research considers the users as an entirety, and it determines the HO moment under the assumption that all of the satellite subpoints (SSP) pass through the centre of the cell. However, when using this scheme, the HO failure ratio (HOFR) would experience great degradation caused by the imbalance between the unified HO moments and the uncertain spatial distribution of users’ (SDU) spatial-temporal variation. This paper proposes a novel HO moment determination method for a low-orbit satellite internet network (LEO-SIN). The rules of SDU variance under SSP motion are first proposed, and they calculate dynamic UE requests within the constraints of the footprint boundary and with SSP motions. Then, we first formulate the problems of multiuser-directed graphs for HO moment determination and prove that it is a nondeterministic polynomial-time(NP) hard problem. An animal survival algorithm based on the Dingo of algorithm (DOA) is proposed to solve the above problems. Multiuser fused directed graphs are first designed to determine HO moments based on the rules of SDU variation and the animal survival algorithm. The simulations show that the proposed method has a better HO performance for LEO-SIN.

This paper proposes a linear companding transform (CT) using either a single inflection point or two inflection points to reduce the peak-to-average power ratio (PAPR) in orthogonal time-frequency space (OTFS) signals. The CT strategically compresses higher amplitudes and enhances lower amplitudes based on carefully chosen scaling factors and points of inflection. With these selected parameters, the CT effectively reduces peak power while maintaining average power, leading to a substantial decrease in PAPR. We analyze noise changes in the inverse companding transform (ICT) process. The analysis reveals that the ICT amplifies less than $20 \%$ of the total noise. A convolutional encoder and soft decision Viterbi decoding algorithm are utilized in the OTFS system to improve the detection performance. We present simulation results focusing on PAPR reduction and bit error rate (BER) performance. These results demonstrate that the CT with two inflection points outperforms both the single inflection point case and the existing $\mu$-law companding, clipping, peak windowing, unique OTFS frame structure, selected mapping, and partial transmit sequence methods, achieving significant PAPR reduction and BER performance.

The integration of satellite communication network and cellular network has a great potential to enable ubiquitous connectivity in future communication networks. Among numerous related application scenarios, the direct connection of mobile phone to satellite has attracted increasing attention. However, the spectrum scarcity in the sub-6 GHz band and low spectrum utilization prevents its popularity. To address these problem, in this paper, we propose a dynamic spectrum sharing method for satellite network and cellular network based on beam-hopping. Specifically, we first develop a centralized dynamic spectrum sharing architecture based on beam-hopping, and propose a delay pre-compensation scheme for beam hopping pattern. Then, an optimization problem is formulated to maximize the overall capacity of the integrated network, with considering the service requirements, the fairness between beam positions and mixed co-channel interference, etc. To solve this problem, a polling-based dynamic resource allocation algorithm is proposed. Simulation results confirm that the proposed algorithm can effectively reduce the serious co-channel interference between different beams or different systems, and improve the spectrum utilization rate as well as system capacity.

Space laser communication (SLC) is an emerging technology to support high-throughput data transmissions in space networks. In this paper, to guarantee the reliability of high-speed SLC links, we aim at practical implementation of low-density parity-check (LDPC) decoding under resource-restricted space platforms. Particularly, due to the supply restriction and cost issues of high-speed on-board devices such as analog-to-digital converters (ADCs), the input of LDPC decoding will be usually constrained by hard-decision channel output. To tackle this challenge, density-evolution-based theoretical analysis is firstly performed to identify the cause of performance degradation in the conventional binary-initialized iterative decoding (BIID) algorithm. Then, a computation-efficient decoding algorithm named multiary-initialized iterative decoding with early termination (MIID-ET) is proposed, which improves the error-correcting performance and computation efficiency by using a reliability-based initialization method and a threshold-based decoding termination rule. Finally, numerical simulations are conducted on example codes of rates 7/8 and 1/2 to evaluate the performance of different LDPC decoding algorithms, where the proposed MIID-ET outperforms the BIID with a coding gain of 0.38 dB and variable node calculation saving of 37%. With this advantage, the proposed MIID-ET can notably reduce LDPC decoder's hardware implementation complexity under the same bit error rate performance, which successfully doubles the total throughput to 10 Gbps on a single-chip FPGA.

Linear programming (LP) decoding is a classic decoding method for linear block codes, and has attracted recent researches because its potential in joint channel processing. However, for polar codes, LP decoders has long been outperformed by CRC-aided successive cancellation list (CA-SCL) decoders. To increase the competitiveness of 5G NR LP polar decoding, it is possible to gain performance improvements by exploiting the cyclic redundancy check (CRC) setup. In this paper, we propose a combined scheme of reduced sparsified factor graph-sparsified CRC (RSFG-SCRC) and augmented generator matrix-CRC (AGM-CRC), for polytope generation in adaptive linear programming (ALP) decoder for 5G polar codes. Augmented generator matrix (AGM) polytope and improved maximum cycle strategy - auxiliary node pairs 4 (MCS-ANP-4) algorithm are proposed, to make efficient use of CRC constraints and minimize the constraint size for the decoder. Numerical simulations show that adaptive linear programming decoders with our proposed RSFG-SCRC and AGM-CRC polytopes can achieve significantly better block error rate (BLER) performance than a benchmark CA-SCL-8 decoder especially in harsh low-to-medium SNR regions.

With the gradual penetration of the internet of things (IoT) into all areas of life, the scale of IoT devices shows an explosive growth trend. The era of internet of everything is coming, and the important position of IoT security is becoming increasingly prominent. Due to the large number types of IoT devices, there may be different security vulnerabilities, and unknown attack forms and virus samples are appear. In other words, large number of IoT devices, large data volumes, and various attack forms pose a big challenge of malicious traffic identification. To solve these problems, this paper proposes a concept drift detection and adaptation (CDDA) method for IoT security framework. The AI model performance is evaluated by verifying the effectiveness of IoT traffic for data drift detection, so as to select the best AI model. The experimental test are given to confirm that the feasibility of the framework and the adaptive method in practice, and the effect on the performance of IoT traffic identification is also verified.

Dual-output power amplifiers (PAs) have shown great potential in the area of radar, satellite and wireless communication systems. However, the flexibility of the power allocation in a dual-output PA without sacrificing efficiency and circuit complexity still needs further investigation. This paper presents a digitally dual-input dual-output (DIDO) PA with reconfigurable modes for power allocation application. The proposed DIDO PA is consist of two identical sub-amplifiers and a 90$^\circ$ coupler, showing a simple circuit topology. The input amplitudes of the two sub-amplifiers and their phase difference is dynamically controlled leveraging on the dual-input technique, leading to reconfigurable operation modes from power allocation to Doherty. In the power allocation mode, flexible power allocation between two output ports can be obtained by the DIDO PA without sacrificing drain efficiency (DE). On the other hand, flexible power transferring and enhanced back-off DE can be simultaneously achieved by the DIDO PA when it is in the Doherty mode. As a proof of concept, a DIDO PA operating at 2.4 GHz is fabricated and measured in this paper. In the power allocation mode, the DIDO PA achieves a DE of more than 71.8% with a total output power of larger than 44 dBm. Moreover, when the DIDO PA operates in the Doherty mode, it could deliver a maximum output power of more than 44 dBm with a saturation DE of more than 73.9% and a 6 dB back-off DE of more than 61.2%.

Temporal link prediction has attracted increasing attention in various fields of complex network analysis, which has important value in the theory and application. However, many existing similarity-based temporal link prediction methods, only analyze the influence of the edge or the point, ignoring the influence of the structures in the network. In this paper, both the spatial-domain model and the time-domain model are taken into consideration, and a novel temporal link prediction method based on the evolution of motif features (TLP-EMF) is proposed. Firstly, a new generalized semi-triangle motif is proposed. And the multi-level contribution of motif point (MP) and motif edge (ME) are described, which is based on the relationship between the full-triangle and the semi-triangle. Secondly, the motif point density (MPD) index and the motif edge density (MED) index are also proposed in a similar way. Thirdly, a novel motif character fusion index (MCF) and a novel motif character density index (MCD) are proposed for the spatial-information processing. Furthermore, a novel forecasting model of the adaptive exponential weighted moving (AEWM) method is proposed for the time-domain evolution. It uses the one-order exponential function to fit the effect of time evolution and uses the global attenuation parameter to adaptively quantify the changes in exponential parameters. Experiments on three real social network data sets show that the proposed method can effectively improve the accuracy of temporal link prediction.

Traditionally, a continuous-wave (CW) signal is used to simulate RF circuits during the design procedure, while the fabricated circuits are measured by modulated signals in the test phase, because modulated signals are used in reality. It is almost impossible to use a CW signal to predict system performances, such as error vector magnitude (EVM), bit error rate (BER), etc., of a transceiver front-end when dealing with complex modulated signals. This paper develops an integrated system evaluation engine (ISEE) to evaluate the system performances of a transceiver front-end or its sub-circuits. This cross-domain simulation platform is based on Matlab, advanced design system (ADS), and Cadence simulators to link the baseband signals and transceiver frond-end. An orthogonal frequency division multiplex (OFDM) modem is implemented in Matlab for evaluating the system performances. The modulated baseband signal from Matlab is dynamically fed into ADS, which includes transceiver front-end for co-simulation. The sub-block circuits of the transceiver front-end can be implemented using ADS and Cadence simulators. After system-level circuit simulation in ADS, the output signal is dynamically delivered to Matlab for demodulation. To simplify the use of the co-simulation platform, a graphical user interface (GUI) is constructed using Matlab. The parameters of the OFDM signals can be easily reconfigured on the GUI to simulate RF circuits with different modulation schemes. To demonstrate the effectiveness of the ISEE, a 3.5 GHz power amplifier is simulated and characterized using 20 MHz 16- and 64-QAM OFDM signals.

With the increasing demand for indoor localization, indoor location based on Wi-Fi has gained wide attention due to its convenience of access. In this paper, we propose a new multi-feature fusion convolutional neural network (CNN) based on channel state information (CSI) images, which contains more feature information by constituting a new CSI image with amplitude and angle of arrival information of CSI information collected at known points. Moreover, the global mean filtering (GMC) algorithm with median filtering proposed in this paper is used to filter and reduce the noise of CSI images to obtain clearer images for network training. To extract more features from the CSI images, the traditional single-channel network is extended, and a two-channel design is introduced to extract feature information between adjacent subcarriers. Experimental evaluation is performed in a typical indoor environment, and the proposed method is experimentally proven to have good localization performance.

This paper studies the use of unmanned aerial vehicles (UAV) equipped with radio frequency (RF) energy harvesting (EH) technology to quickly establish temporary communication networks in disaster areas to provide artificial intelligence (AI) based services to users with limited resources. In particular, to ensure the quality of AI-based services and improve the lifetime of emergency communication networks, we study how to reduce the service latency and energy consumption when fine-tuning models of AI-based services in the resource-constrained emergency system. A joint optimization problem of model training and RF EH for UAV-based emergency communication network is formulated. Due to the nonlinear RF EH circuit characteristics, the optimization problem is non-convex. We transform the non-convex problem into solvable subproblems and propose an energy-efficient and low-latency federated learning algorithm (EL-FL) to solve these subproblems. Theoretical analysis of the convergence and computational complexity of EL-FL is provided. Simulation results show that the proposed scheme significantly outperforms other baseline methods in various network environments.

In 6G, artificial intelligence represented by deep nerual network (DNN) will unleash its potential and empower IoT applications to transform into intelligent IoT applications. However, whole DNN-based inference is difficult to carry out on resource-constrained intelligent IoT devices and will suffer privacy leakage when offloading to the cloud or mobile edge computation server (MECs). In this paper, we formulate a privacy and delay dual-driven device-edge collaborative inference (P4DE-CI) system to preserve the privacy of raw data while accelerating the intelligent inference process, where the intelligent IoT devices run the front-end part of DNN model and the MECs execute the back-end part of DNN model. Considering three typical privacy leakage models and the end-to-end delay of collaborative DNN-based inference, we define a novel intelligent inference Quality of service (I2-QoS) metric as the weighted summation of the inference latency and privacy preservation level. Moreover, we propose a DDPG-based joint DNN model optimization and resource allocation algorithm to maximize I2-QoS, by optimizing the association relationship between intelligent IoT devices and MECs, the DNN model placement decision, and the DNN model partition decision. Experiments carried out on the AlexNet model reveal that the proposed algorithm has better performance in both privacy-preserving and inference-acceleration.

In this paper, we propose a rate splitting multiple access (RSMA) based integrated sensing and communication system (ISAC), where the sensing and communication are realized simultaneously with the RSMA signal. Further, reconfigurable holographic surface (RHS) is utilized to replace the traditional antennas for beam generation, expecting to combine the advantages of RSMA and RHS. To maximize the weighted summation of system rate and probing power, an optimization problem is formulated to jointly design the digital beamformer, the holographic beamformer and the message splitting vectors. To solve the non-convex problem, we first decompose it into two subproblems, where one jointly designs the digital beamformer and message splitting vectors, and the other deals with the holographic beamformer. An iterative algorithm, which leverages successive convex approximation and semi-definite relaxation, is proposed to achieve the sub-optimal solution through solving these two subproblems alternatively. Simulations confirm the effectiveness and efficiency of the proposed algorithm.

Deep reinforcement learning is broadly employed in the optimization of wireless video transmissions. Nevertheless, the instability of the deep reinforcement learning algorithm affects the further improvement of the video transmission quality. The federated learning method based on distributed data sets was used to reduce network costs and increase the learning efficiency of the deep learning network model. It solved too much data transfer costs and broke down the data silos. Intra-clustered dynamic federated deep reinforcement learning (IcD-FDRL) was constructed in clustered mobile edge-computing (CMEC) networks due to the promoted video transmission quality for the stability and efficiency of the DRL algorithm. Then, the IcD-FDRL algorithm was employed to CMEC networks’ edge for intelligent-edge video transmissions, which could satisfy the diversified needs of different users. The simulation analysis proved the effectiveness of IcD-FDRL in improving QoE, cache hit ratio, and training.

Benefitting from UAVs' characteristics of flexible deployment and controllable movement in 3D space, odor source localization with multiple UAVs has been a hot research area in recent years. Considering the limited resources and insufficient battery capacities of UAVs, it is necessary to fast locate the odor source with low-complexity computation and minimal interaction under complicated environmental states. To this end, we propose a multi-UAV collaboration based odor source localization (MUC-OSL) method, where source estimation and UAV navigation are iteratively performed, aiming to accelerate the searching process and reduce the resource consumption of UAVs. Specifically, in the source estimation phase, we present a collaborative particle filter algorithm on the basis of UAVs' cognitive difference and collaborative information to improve source estimation accuracy. In the following navigation phase, an adaptive path planning algorithm is designed based on partially observable Markov decision process to distributedly determine the subsequent flying direction and moving steps of each UAV. The results of experiments conducted on two simulation platforms demonstrate that MUC-OSL outperforms existing efforts in terms of mean search time and success rate, and effectively reduces the resource consumption of UAVs.

Low earth orbit (LEO) satellite networks can provide wider service coverage and lower latency than traditional terrestrial networks, which have attracted considerable attention. However, the uneven distribution of human population and data traffic on the ground incurs unbalanced traffic load in LEO satellite networks. To this end, we propose a load-balancing routing algorithm for LEO satellite networks based on ant colony optimization and reinforcement learning. In the ant colony algorithm, we improve the pheromone update rule by introducing load-aware heuristic information, e.g., the current node transmission overhead, delay and load status, and reinforcement learning-based link quality evaluation. It enables the routing algorithm to select the lightly loaded node as the next hop to balance the network load. We simulate and verify the proposed algorithm using the NS2 simulation platform, and the results show that our algorithm improves the data delivery ratio and throughput while ensuring lower latency and transmission overhead.

An orderly competition mechanism is used to change unexpected competition into predictable competition so as to reduce access collision during access process. The scheme is realized by learning, queuing, and accessing. Queuing is the key step to reduce random and realize orderly competition. Related parameters leading to access random including the arrival rate, the delay requirements, the number of devices, and so on, are defined as queue factors in this paper. The queue factors are obtained from the improved double deep Q network (DDQN) algorithm which is proposed here by setting asynchronous weights of two target networks. By learning, the queue factors will guide the devices with diverse delay requirements to queue. Then the queued devices start the access process according to their learning optimal access slot and preamble. Different from traditional competition solutions, markov decision process of the orderly competition mechanism has only two states, which remarkably cuts down the back-off rate and reduces the access delay. The simulation results show that the access success rate of this method can be close to 100% before the system capacity approaches the maximum value.

This paper introduces a framework aimed at aiding the development of sixth-generation (6G) ultra-massive machine type communications (um-MTC). Precisely, the deployment of wireless power transfer (WPT) supported device-to-device (D2D) communication occurs within multiple-input single-output non-orthogonal multiple access (MISO-NOMA) downlink networks to facilitate spectrum and energy collaboration. A pure fractional programming (PFP) algorithm is proposed to maximize the WPT-assisted device's energy efficiency. An optimal closed-form solution for determining the time-switching coefficient of the WPT device is provided. For the robust beamforming design, the complex multi-dimension quadratic transform is applied. Moreover, the paper applies the deep deterministic policy gradient (DDPG)-based approach to directly address the problem and compares it with the proposed algorithm. Simulation outcomes highlight two key insights: 1) The PFP algorithm surpasses the performance of the DRL-based algorithm when the acquired channel state information (CSI) is accurate or contains negligible errors, while the opposite is true for imperfect CSI 2) The higher energy efficiency gains can be achieved in NOMA scheme than that in Orthogonal Multiple Access (OMA) scheme.

To meet the increasing demands for wideband communications of the vehicle's infotainment applications such as the virtual reality (VR), the millimeter wave (mmWave) enabled connected automated vehicles (CAVs) is of great demand. However, the mmWave vehicular communication brings new challenges on the content distribution efficiency in terms of the differentiated VR's service requirements and the dynamic interference among vehicle-to-infrastructure (V2I) and vehicle-to-vehicle (V2V) links. Therefore, this paper proposes an interference cognition based mmWave beam resource allocation algorithm for CAVs to maximize the content distribution efficiency and minimize the interference among CAVs. In the V2I stage, the interference prediction assisted V2I vehicle selection algorithm is proposed, which can aware the intra base station (BS) interference dynamically. Moreover, the coalition game based V2V content distribution algorithm is proposed, where a novel cache-hit and interference aware utility function is designed. Simulation results prove that the average successful transmission probability of the proposed algorithm can reach $72.63\%$, which is $53.4\%$ higher than the conventional algorithms.

In this paper, we study the power allocation problem in energy harvesting internet of things (IoT) communication system, with the aim to maximize the total throughput while avoiding data buffer overflow or energy exhausting. The IoT node has a finite battery to store the harvested energy and a limited buffer for the storage of the unsent data. The energy/data arrives following a Markov process. Assuming the node has no prior knowledge of the energy/data process and only knows the values of the current time slot, the optimal power allocation problem is modeled as a reinforcement learning task. The state consists of the data in the buffer, the energy stored in the battery, the new coming data amount, the energy harvesting amount and the channel coefficient at time slot $t$. Then the action is defined as the selected transmitting power. With the growth of the state or action space, it is challenging to visit every state-action pair sufficiently and store all the state-action values, so a deep Q-learning based algorithm is proposed to solve this problem. Simulation results show the advantages of our proposed algorithms, and we also analyze the effect of different system setting parameters.