With the rapid proliferation of Internet of Things (IoT) devices and systems, the demand for secure and efficient data trading between networks is growing. Blockchain offers a trusted solution with its inherent decentralization and tamper-proof characteristics. However, heterogeneous blockchains used by different IoT networks lead to difficulties in inter-chain data exchange and further form a "data silo". To resolve these issues, this paper proposes a cross-chain collaborative data transfer channel based on asynchronous confirmation transaction stacks (CCTS) specifically designed for IoT environments’ data trading. By leveraging a relay chain architecture, our solution enables secure cross-chain transaction negotiation and coordinates efficient off-chain data transmission between IoT devices and systems, achieving anti-repudiation and traceability for critical IoT data exchanges. Experimental results demonstrate that the method supports millisecond-to-second latency for cross-chain transactions and maintains low on-chain storage overhead suitable for resource-constrained IoT nodes. Overall, CCTS provides a scalable framework for building robust, interoperable data trading networks with verifiable data provenance.

Toxic content detection is crucial for online services to remove inappropriate content that violates community standards. To automate the detection process, prior works have proposed varieties of machine learning (ML) approaches to train Language Models (LMs) for toxic content detection. However, both their accuracy and transferability across datasets are limited. Recently, Large Language Models (LLMs) have shown promise in toxic content detection due to their superior zero-shot and few-shot in-context learning ability as well as broad transferability on ML tasks. However, efficiently designing prompts for LLMs remains challenging. Moreover, the high run-time cost of LLMs may hinder their deployments in production. To address these challenges, in this work, we propose BD-LLM, a novel and efficient approach to bootstrapping and distilling LLMs for toxic content detection. Specifically, we design a novel prompting method named Decision-Tree-of-Thought (DToT) to bootstrap LLMs' detection performance and extract high-quality rationales. DToT can automatically select more fine-grained context to re-prompt LLMs when their responses lack confidence. Additionally, we use the rationales extracted via DToT to fine-tune student LMs. Our experimental results on various datasets demonstrate that DToT can improve the accuracy of LLMs by up to 4.6%. Furthermore, student LMs fine-tuned with rationales extracted via DToT outperform baselines on all datasets with up to 16.9% accuracy improvement, while being more than 60x smaller than conventional LLMs. Finally, we observe that student LMs fine-tuned with rationales exhibit better cross-dataset transferability.

In parallel computing, Data Flow Graphs play a critical role by explicitly representing the dependencies between tasks, which is essential for task scheduling and resource utilization. In the process of scheduling optimization, it is crucial to address the dual challenges of balancing computational load and distributing data access pressure. The Dataflow for C++ (DFCPP) possesses the advantage of accurately sensing task data sizes, and based on this capability, this paper proposes a Primary-Secondary Core Selecting (PSCS) strategy specifically designed for Hyper-Threading-enabled Non-Uniform Memory Access systems (HT-NUMA). This strategy aims to mitigate interference between threads on the same physical core in a hyper-threading environment, thereby enhancing task execution stability and overall performance. Additionally, DFCPP integrates a task-stealing mechanism with a proactive allocation strategy for large-cache tasks, enabling the distribution of large tasks to low-load NUMA nodes, which reduces cache contention and improves cache hit rates. Experimental results demonstrate that, compared to common dataflow programming libraries such as Taskflow and TBB, DFCPP achieves over a 20% performance improvement when the system core count meets computational demands, and nearly a 25% improvement when handling large-dataset tasks. Furthermore, DFCPP exhibits high efficiency and adaptability in processing tasks with various directed acyclic graph topologies, fully leveraging the parallel computing advantages of NUMA systems, and demonstrates exceptional performance and broad applicability in dataflow task scheduling, resource optimization, and complex dependency management.

Atmospheric weighted mean temperature (Tm) is a crucial parameter that links precipitable water vapor (PWV) and zenith wet delay (ZWD). To address the challenge of balancing the quality and timeliness of Tm data, this study introduced infrared remote-sensing technology based on meteorological satellites for the first time to retrieve Tm. We developed separate Tm estimation models for FengYun-4A (FY4A) observations under both clear and cloudy conditions and combined them to enable real-time retrieval of all-weather Tm. This combined model, called the all-weather Tm estimation model, is based on the linear relationship between Tm and surface temperature as well as remote sensing retrieval theories related to surface temperature and cloud-top properties. This grouping modeling approach allows continuous spatiotemporal Tm data to be estimated at minute intervals, even under cloudy conditions. Radiosonde-derived and ERA5-derived Tm data from 2022 were used to assess the accuracy of FY4A-derived Tm for Australia. Compared to radiosonde-derived Tm, the root mean square error (RMSE)/bias values for FY4A-derived Tm were 1.37/0.05, 1.45/0.06, and 1.38/0.06 K for all-time, daytime, and nighttime, respectively. Compared to the ERA5-derived Tm, the RMSE/bias values for FY4A-derived Tm were 1.26/0.01, 1.33/0.01, and 1.37/0.03 K under all-weather, clear, and cloudy conditions, respectively. The validation results indicated that the satellite-based Tm retrieval model possesses the advantages of real-time monitoring, all-weather capability, high accuracy, and high spatiotemporal resolution. Thus, it has tremendous potential for deepening interdisciplinary collaboration between the meteorology and navigation fields.

Practical natural language processing (NLP) tasks often exhibit long-tailed distributions accompanied by noisy labels, posing significant challenges to the generalization and robustness of complex models such as Deep Neural Networks (DNNs). Traditional resampling techniques like oversampling or undersampling, while commonly employed, can easily lead to overfitting. A growing trend involves leveraging small amounts of metadata to learn data weights, alongside the demonstrated benefits of self-supervised pre-training, particularly for under-represented data. In this work, we propose a general framework that addresses both the long-tail and noisy label issues. Our model is adaptively tailored to the problem domain using a contrastive learning approach. The re-weighting module, a feed-forward network, learns explicit weighting functions and adjusts weights based on metadata. Additionally, our framework modifies the weights of terms in the loss function through a combination of polynomial expansion of the cross-entropy loss and focal loss. Extensive experiments consistently demonstrate the superior performance of our proposed framework compared to baseline methods. Finally, our comprehensive sensitivity analysis underscores the efficacy of the proposed framework in handling long-tailed problems and mitigating the adverse effects of noisy labels.

The latest batch of the Chinese Fengyun-2 (FY-2) geostationary satellites (i.e., FY-2F, FY-2G, and FY-2H) provides total precipitable water (TPW) products at high spatial and temporal resolutions. However, due to the lack of accuracy and performance evaluation for these products, a vast amount of valuable TPW data remains unused in atmospheric science research. To address this issue, this study aimed to propose and apply a validation strategy that incorporated a hemispheric vertical correction model (VCM) to obtain reliable evaluation results. With the help of reliable radiosonde and the state-of-the-art fifth generation of the European Center for Medium-Range Weather Forecasts (ECMWF) reanalysis data (ERA5), this study was the first to assess the quality of the full-disk TPW products retrieved via the three FY-2 satellites from January 2019 to December 2020. In addition, this study analyzed the water vapor content, latitude, and elevation dependencies of FY-2 TPW retrieval error and explored the potential for improving the quality of each satellite TPW product through a linear calibration in the three test areas of northern and southern temperate zones and tropics. The results of this study were threefold. First, the accuracy of FY-2F and FY-2H TPW was superior to that of FY-2G. The root mean square error (RMSE) values of FY-2F, FY-2G, and FY-2H were 3.84, 4.46, and 3.73 mm and 2.30, 2.55, and 2.14 mm relative to the radiosonde and ERA5 data, respectively. Second, the TPW retrieval error of the FY-2 satellites decreased with the increasing latitude or decreasing elevation. Overall, FY-2G underestimated TPW, whereas FY-2F and FY-2H only underestimated TPW under wet conditions (i.e., TPW > 55 mm). Finally, the calibration potential of FY-2F and FY-2H was higher than that of FY-2G, and the bias, slope, and potential index (PI) values of FY-2G were lower than those of FY-2H and FY-2F in all the three test areas.

This paper investigates the attack detection problem for linear parameter-varying (LPV) based-model Cyber-physical system (CPS). Considering the time-varying delay, limited communication, fault signal, and the actuator cyber-attack, an H ∞ observer is constructed, so that the system can detect the attack signal. On the basis of the parameter-dependent Lyapunov-Krasovskii functional and the linear matrix inequality (LMI) approach, adequate criteria are provided that ensure the system satisfies the asymptotic stability and H ∞ performance. Moreover, the infinite-dimensional feasibility criteria are converted to a finite-dimensional set of LMIs by using the basic functions and grid technology. Finally, a simulated example verifies the validity of the offered attack detection method.

Precipitable water vapor (PWV) is one of the key parameters in the evolution of extreme weather and climate change. However, current data fusion methods (such as Gaussian processes, spherical cap harmonics, and polynomial fitting) can hardly obtain simultaneously the PWV map with high precision and high spatiotemporal resolution. To solve this problem, a two-step-based PWV fusion (TPF) method is proposed, in which a hybrid PWV fusion model (HPFM) and a spatial and temporal fusion model (STFM) are introduced separately. In the first step, HPFM is established by combining the global pressure and temperature 2 wet (GPT2w) model, spherical harmonic functions, and polynomial fitting to obtain the PWV value with high precision at an arbitrary location in the study area. In the second step, STFM is proposed to generate the PWV map with high temporal resolution taking advantage of site-based global navigation satellite system (GNSS)-derived PWV. To validate the performance of the proposed method, GNSS observations, ERA-Interim, and ERA5 reanalysis products are selected in Yunnan Province, China, to carry out the experiment. Statistical results show that: 1) HPFM has the ability to obtain atmospheric water vapor with a root mean square (rms) of less than 3 mm in an arbitrary location of the PWV map and 2) STFM can generate PWV maps with the same temporal resolution as GNSS observations, and the accuracy of the obtained PWV values can be guaranteed. Therefore, the proposed TPF method is proven to have the ability to simultaneously retrieve PWV maps with high accuracy and spatiotemporal resolution.

Spoken language understanding (SLU) is the task of inferring the semantics of spoken utterances. Traditionally, this has been achieved with a cascading combination of Automatic Speech Recognition (ASR) and Natural Language Understanding (NLU) modules that are optimized separately, which can lead to a suboptimal overall performance. More recently, End-to-End SLU (E2E SLU) was proposed to perform SLU directly from speech through a joint optimization of the modules, addressing some of the traditional SLU shortcomings. A key challenge of this approach is how to best integrate the feature learning of the ASR and NLU sub-tasks to maximize their performance. While it is known that in general, ASR models focus on low-level features, and NLU models need higher-level contextual information, ASR models can nonetheless also leverage top-down syntactic and semantic information to improve their recognition. Based on this insight, we propose Top-Down SLU (TD-SLU), a new transformer-based E2E SLU model that uses top-down attention and an attention gate to fuse high-level NLU features with low-level ASR features, which leads to a better optimization of both tasks. We have validated our model using the public FluentSpeech set, and a large custom dataset. Results show TD-SLU is able to outperform selected baselines both in terms of ASR and NLU quality metrics, and suggest that the added syntactic and semantic high-level information can improve the model’s performance.

In this paper, the H−/H∞detection for faults and attacks in a Cyber-physical system (CPS) is studied. The complex network environment with random delay, limited communication, actuator cyber attack and sensor cyber attack is considered. And the above system with fault and attack coexisting is discretized by delta operator in high-speed sampling. A set of H−/H∞observers is constructed, which enables the system to detect the possible fault signal and attack signal respectively and quickly. By using Lyapunov-Krasovskii functional and linear matrix inequalities approach, the sufficient conditions for the system to have asymptotic stability and H−/H∞performance are given. The feasibility and effectiveness of the proposed method are analyzed through several simulation examples.

In this paper, we demonstrate a new dataflow platform of DFC, which can handle the successive dataflow computing passes with tagged data. By implementing the matrix multiplication in DFC, we show that DFC can exploit the parallelism automatically with a much simple dataflow graph constructed by DF functions of DFC. Different from the other dataflow execution platform, DFC support multiple worker threads for one dataflow node of DF functions. By running the matrix multiplication program of DFC on Kunlun system, it was verified that DFC get a reasonable speedup for large scale computing for thread number up to 512.

Hash table is used in many areas of networking such as route lookup, packet classification, per-flow state management and network monitoring for its constant access time latency at moderate loads. However, collisions may become frequent at high loads in traditional hash tables, which may lead the access time complexity to be linear and intolerable to applications like high-speed route lookups. While some schemes were proposed to help resolve this problem and most of them may achieve O(1) average memory access per lookup, very few of them are able to cut down the access to a deterministic single one. In this paper, we design a structure called deterministic and efficient hash table (DEHT). In DEHT, a novel data structure on on-chip memory is built, with the help of which the off-chip memory access can be decreased to a single one at most per lookup even when the load of the hash table is very high. What's more, the on-chip data structure also plays a similar role as Bloom Filter to do membership screening, which can avoid most lookups of nonexistent items of the hash table visiting the off-chip memory. Through theoretical analysis and simulations, we show that our scheme is faster than other schemes in lookup operations; the usable load of the off-chip hash table, the memory efficiency and the false positive rate of the on-chip data structure are also favorable.

TCP Incast is one of the typical datacenter problems that may affect network efficiency in a many to one transmission session, in which the completion time of the whole session depends on the last finishing flow. In this paper a session-aware mechanism is proposed that intelligently chooses only the leading flows to slow down at the presence of congestion, so the lagging ones will have a higher chance to catch up to achieve better session completion time and goodput. Compared to existing solutions, the proposed one introduces no change to the TCP protocol and no extra messaging for the internal status of either TCP or the intermediate switches. Experimental results show that the proposed mechanism can indeed balance the progress among Incast flows and achieve a shorter session time and higher goodput.

In Named Data Networking (NDN), the speed of name lookup operation and the scalability concern impact deeply on the performance of the forwarding system. NDN names can contain unbounded number of components which may result in intolerable lookup time. And since routers cannot have unlimited table size to keep up with the unbounded nature of application data namespace of NDN, it may lead to missing records for valid data names in the routing table. In this paper, we propose a port information assisted trie or P-trie to help solve these two issues: name lookup process is accelerated with the help of the port information on nodes of the trie, and packets with names not in the routing tables are forwarded to a most promising port based on the analysis of existing entries. We implemented P-trie in a multi-aligned transition array (MATA) to reduce the memory footprint. Experiments show that the name lookup rate of our P-trie can be times faster than other trie based methods.

Coherent modulation is more effective than differential modulation for orthogonal frequency division multiplexing (OFDM) systems requiring high data rate and spectral efficiency. Channel estimation is therefore an integral part of the receiver design. Two iterative maximum likelihood (ML) based channel estimation algorithms are proposed for OFDM uplinks in dispersive time-varying channels. The uplink multipath fading channel is modeled such that the channel state can be determined by estimating the unknown channel parameters. A second-order Taylor series expansion is adopted to simplify the channel estimation problem. Based on the system model, an iterative ML-based algorithm is first proposed to estimate the discrete-time channel parameters. The mean square error performance of the proposed algorithm is analyzed using a small perturbation technique. Based on a convergence rate analysis, an improved iterative ML channel estimation algorithm is presented using a successive overrelaxation method. Numerical experiments are performed to confirm the theoretical analyses and show the improvement in convergence rate of the improved algorithm.

Maximum-likelihood based channel estimation is considered for orthogonal frequency division multiplexing systems in dispersive time-varying channels. A successive overrelaxation approach is adopted to analyze the convergence rate of an iterative channel estimation algorithm. Based on the analysis, an improved fast converging iterative channel estimation algorithm is proposed. Numerical tests are performed to show the improvement of the proposed algorithm over a recently proposed channel estimation algorithm.

Asymptotic error rate expressions are derived for equal gain combining and selection combining on dual-branch correlated Rician channels. The asymptotic error rate performances of three popular diversity combining techniques are compared and some insights are obtained. The impacts of various channel parameters on the asymptotic performances of these diversity combining techniques are studied comprehensively. Some previous observations on diversity performances on correlated Rician channels are unified and further extended.

This paper presents a spectral band selection method for feature dimensionality reduction in hyperspectral image analysis for detecting skin tumors on poultry carcasses. A hyperspectral image contains spatial information measured as a sequence of individual wavelength across broad spectral bands. Despite the useful information for skin tumor detection, real-time processing of hyperspectral images is often a challenging task due to the large amount of data. Band selection finds a subset of significant spectral bands in terms of information content for dimensionality reduction. This paper presents a band selection method of hyperspectral images based on the recursive divergence for the automatic detection of poultry carcasses. For this, we derive a set of recursive equations for the fast calculation of divergence with an additional band to overcome the computational restrictions in real-time processing. A support vector machine is used as a classifier for tumor detection. From our experiments, the proposed band selection method shows high detection accuracy with low false positive rates compared to the canonical analysis at a small number of spectral bands. Also, compared with the enumeration approach of 93.75% detection rate, our proposed recursive divergence approach gives 90.6% detection rate, which is within the industry-accepted accuracy of 90–95%, while achieving the computational saving for real-time processing.

The bit error rate of a binary phase shift keying (BPSK) signal in Ricean-faded cochannel interference is studied. A precise bit error rate expression is derived for a bandlimited BPSK signal corrupted by an arbitrary number of asynchronous Ricean-faded interfering signals. An error rate estimation based on a saddlepoint approximation is also provided. It is shown that this approximation is highly accurate. For the special case, when there is one synchronous Ricean-faded interfering signal in an interference-limited environment, an asymptotic error rate analysis based on the saddle point approximation reveals that fading in the cochannel interfering signal can have different impacts on the desired user's error rate performance. Our study shows that when the signal-to-interference ratio (SIR) is sufficiently large, the error floor of the desired user signal decreases with an increase of the Rice factor in the interfering user's fading channel. However, the opposite phenomenon can happen when the SIR is small.

An asymptotic error rate performance analysis has been reported recently for maximum ratio combining on correlated Rician channels; however, a similar approach cannot be applied to equal gain combining or selection combining. A novel approach is introduced to analyze the asymptotic error rate performances of equal gain combining and selection combining on dual-branch correlated Rician channels. The asymptotic error rate performances of popular diversity combining techniques are compared and some insights are obtained that explain some behaviors previously considered unexpected. The impacts of various channel parameters on the asymptotic performances of these diversity combining techniques are studied comprehensively. Some previous observations on diversity performances on correlated Rician channels are unified and further extended.

Time variation on fading channels hinders accurate channel estimation in differential space-time modulation and deteriorates the performance. Decision-feedback differential detection is studied for block differential space-time modulation, and compared with conventional differential space-time modulation. It is observed that the proposed scheme does not suffer effective fading bandwidth expansion, as does the conventional scheme. An improved effective signal-to-noise ratio approach is proposed for analyzing the performance of the proposed scheme in time-varying flat Rayleigh fading. Theoretical analysis and simulations show the improved performance of the proposed scheme over the conventional scheme.

Error rates of orthogonal frequency-division multiplexing (OFDM) signals in multipath slow fading Nakagami-m fading channels are considered. The exact probability density function of a sum of Nakagami-m random phase vectors is used to derive a closed-form expression for the error rates of OFDM signals. The precise error-rate analysis is extended to a system using multichannel reception with maximal ratio combining. An asymptotic error-rate analysis is also provided. For a two-tap channel with finite values of Nakagami-m fading parameters, our analysis and numerical results show that the asymptotic error-rate performance of an OFDM signal is similar to that of a single carrier signal transmitted over a Rayleigh fading channel. On the other hand, our analysis further shows that a frequency-selective channel that can be represented by two constant taps has similar asymptotic error-rate performance to that of a one-sided Gaussian fading channel. It is observed that, depending on the number of channel taps, the error-rate performance does not necessarily improve with increasing Nakagami-m fading parameters.

Time variation on fading channels hinders accurate channel estimation in differential space-time modulation and deteriorates the performance. Decision-feedback differential detection is considered for block differential space-time modulation and compared with conventional differential space-time modulation. It is observed that the proposed scheme does not suffer effective fading bandwidth expansion as does the conventional scheme. An improved effective signal-to-noise ratio approach is proposed for analyzing the performance of the system. Theoretical analysis and simulation show the improved performance of the proposed scheme over the conventional scheme.

A new scheme, selective time domain filtering, that has significantly reduced complexity compared to the SLM scheme, which only relies on conventional channel estimation and demodulation techniques for multipath fading channels to recover the data, is proposed. Computer simulations indicate that the new scheme suffers only a small loss in performance in exchange for significant reduction in complexity.

The bit error rate of a binary phase shift keying signal in Ricean-faded cochannel interference is studied. A precise bit error rate expression based on a characteristic function method is derived for a bandlimited binary phase shift keying signal corrupted by an arbitrary number of asynchronous Ricean-faded interfering signals. For the special case when there is one synchronous Ricean-faded interfering signal, a Chernoff bound analysis is performed and it predicts that the error floor of the desired user signal decreases with an increase of the Rice factor in the interfering user's fading channel. However, our precise bit error rate analysis results reveal that the opposite phenomenon can also happen, in particular when the signal-to-interference power ratio is low. A saddle-point approximation based error rate analysis is also provided. It is shown that this approximation is highly accurate. An asymptotic analysis based on the saddle-point approximation further reveals that a minimum signal-to-interference power ratio is required to have the desired user's error rate performance improved by a less-faded interfering signal.

Differential unitary space-time modulation is an attractive space-time technique because it does not require channel estimation. In this paper, we propose a new differential space-time code with rate 2.5 bit/s/Hz for two transmit antennas. In contrast to the original unitary space-time codes with one level of amplitude, the powers of the matrices in this new differential space-time code are different. We design a decoding algorithm based on the Viterbi algorithm to differentially detect the transmitted data. Simulation results show that the new two amplitude level differential space-time code outperforms the original unitary space-time code with rate 2.5 bit/s/Hz.

A new class of differential unitary space-time codes, namely, differential monomial space-time codes, is proposed. In contrast to many other differential unitary space-time codes, this new class of differential monomial space-time codes does not require the transmitted matrices to form a group. Instead, the signals at each transmit antenna are constrained in MPSK constellations. Due to this feature, the transmitted signals on each transmit antenna can be generated by a simple table lookup and modulus-add operation with low complexity. We give design examples for 4 transmit antennas and show the performance improvement over conventional differential unitary space-time codes.

A closed-form expression for the pairwise error probability of cyclic unitary space-time codes with t=2 and r=1 is obtained. Additionally, this result is used to obtain an upper bound to the symbol error rate of cyclic unitary differential space-time group codes. It is also shown that these techniques can be used to approximate the symbol error rate of dicyclic unitary differential space-time group codes.

The error rates of orthogonal frequency division multiplexing (OFDM) signals on frequency selective Nakagami-m fading channels are considered. The exact probability density function of a sum of Nakagami-m random vectors is used to derive a closed-form expression for the error rate of OFDM signals. The exact error rate analysis is also extended to a system using multichannel reception with maximal ratio combining. Our analysis and numerical results show that the error rates obtained using a previous Nakagami-m approximation can be unreliable. It is observed that, depending on the number of channel taps, the error rate performance may degrade with increasing values of Nakagami fading parameters.

OFDM (orthogonal frequency division modulation) is a very promising technique for future communication system. However, frequency offset is well known as one of the main factors that influence the performance of OFDM communication systems. Various coarse frequency synchronization algorithms based on differential detection have been proposed to cope with the oscillation in frequency response in dispersive channel. This paper extends the concept of differential detection and improves existing coarse frequency offset synchronization algorithm. In this modified algorithm, parameter can be adjusted to obtain different performance tradeoff. In addition, compared with the conventional algorithm, the computation complexity of this proposed algorithm only increases a little and remains the same order. Simulations are carried out to verify the improvement of performance of our proposed algorithm.

A linear recursive "one-shot" decorrelator for asynchronous CDMA systems is developed. It has a lower complexity and better performance than existing "one-shot" decorrelators while eliminating all multiple access interference (MAI). This decorrelator does not exploit any signal energy information. Through simulation, this decorrelator is shown to be near-far resistant in both AWGN and fading channel.