In this paper, we propose an online probabilistic activation/deactivation control method for base stations (BSs) in heterogeneous networks based on the temporal system throughput and activation states of neighbor BSs (cells). The conventional method iteratively updates the activation/deactivation states in a probabilistic manner at each BS based on the change in the observed system throughput and activation/deactivation states of that BS between past multiple consecutive discrete times. Since BS activation control increases the system throughput by improving the tradeoff between the reduction in inter-cell interference and the traffic off-loading effect, the activation of a BS whose neighbor BSs are deactivated is likely to result in improved system performance and vice versa. The proposed method newly introduces a metric, which represents the effective ratio of the activated neighbor BSs considering their transmission power and distance to the BS of interest, to the update control of the activation probability. This improves both the convergence rate of the iterative algorithm and throughput performance after convergence. Computer simulation results, in which the mobility of the user terminals is taken into account, show the effectiveness of the proposed method.

Aggregate signature schemes enable us to aggregate multiple signatures into a single short signature. One of its typical applications is sensor networks, where a large number of users and devices measure their environments, create signatures to ensure the integrity of the measurements, and transmit their signed data. However, if an invalid signature is mixed into aggregation, the aggregate signature becomes invalid, thus if an aggregate signature is invalid, it is necessary to identify the invalid signature. Furthermore, we need to deal with a situation where an invalid sensor generates invalid signatures probabilistically. In this paper, we introduce a model of aggregate signature schemes with interactive tracing functionality that captures such a situation, and define its functional and security requirements and propose aggregate signature schemes that can identify all rogue sensors. More concretely, based on the idea of Dynamic Traitor Tracing, we can trace rogue sensors dynamically and incrementally, and eventually identify all rogue sensors of generating invalid signatures even if the rogue sensors adaptively collude. In addition, the efficiency of our proposed method is also sufficiently practical.

Deep neural networks show good performance in image recognition, speech recognition, and pattern analysis. However, deep neural networks also have weaknesses, one of which is vulnerability to poisoning attacks. A poisoning attack reduces the accuracy of a model by training the model on malicious data. A number of studies have been conducted on such poisoning attacks. The existing type of poisoning attack causes misrecognition by one classifier. In certain situations, however, it is necessary for multiple models to misrecognize certain data as different specific classes. For example, if there are enemy autonomous vehicles A, B, and C, a poisoning attack could mislead A to turn to the left, B to stop, and C to turn to the right simply by using a traffic sign. In this paper, we propose a multi-targeted poisoning attack method that causes each of several models to misrecognize certain data as a different target class. This study used MNIST and CIFAR10 as datasets and Tensorflow as a machine learning library. The experimental results show that the proposed scheme has a 100% average attack success rate on MNIST and CIFAR10 when malicious data accounting for 5% of the training dataset have been used for training.

While the introduction of softwarelization technologies such as software-defined networking and network function virtualization transfers the main focus of network management from hardware to software, network operators still have to deal with various and numerous network and computing equipment located in network centers. Toward fully automated network management, we believe that a robotic approach will be essential, meaning that physical robots will handle network-facility management works on behalf of humans. This paper focuses on robotic assistance for on-site network maintenance works. Currently, for many network operators, some network maintenance works (e.g., hardware check, hardware installation/replacement, high-impact update of software, etc.) are outsourced to computing and network vendors. Attendance (witness work) at the on-site vendor's works is one of the major tasks of network operators. Network operators confirm the work progress for human error prevention and safety improvement. In order to reduce the burden of this, we propose three essential works of robots, namely delegated attendance at on-site meetings, progress check by periodical patrol, and remote monitoring, which support the various forms of attendance. The paper presents our implementation of enabling these forms of support, and reports the results of experiments conducted in a commercial network center.

Deep neural networks (DNNs) provide excellent services in machine learning tasks such as image recognition, speech recognition, pattern recognition, and intrusion detection. However, an adversarial example created by adding a little noise to the original data can result in misclassification by the DNN and the human eye cannot tell the difference from the original data. For example, if an attacker creates a modified right-turn traffic sign that is incorrectly categorized by a DNN, an autonomous vehicle with the DNN will incorrectly classify the modified right-turn traffic sign as a U-Turn sign, while a human will correctly classify that changed sign as right turn sign. Such an adversarial example is a serious threat to a DNN. Recently, an adversarial example with multiple targets was introduced that causes misclassification by multiple models within each target class using a single modified image. However, it has the weakness that as the number of target models increases, the overall attack success rate decreases. Therefore, if there are multiple models that the attacker wishes to attack, the attacker must control the attack success rate for each model by considering the attack priority for each model. In this paper, we propose a priority adversarial example that considers the attack priority for each model in cases targeting multiple models. The proposed method controls the attack success rate for each model by adjusting the weight of the attack function in the generation process while maintaining minimal distortion. We used MNIST and CIFAR10 as data sets and Tensorflow as machine learning library. Experimental results show that the proposed method can control the attack success rate for each model by considering each model's attack priority while maintaining minimal distortion (average 3.95 and 2.45 with MNIST for targeted and untargeted attacks, respectively, and average 51.95 and 44.45 with CIFAR10 for targeted and untargeted attacks, respectively).

Wireless sensor networks provide long-term monitoring of the environment, but sensors are powered by small batteries. Using a mobile charger (MC) to replenish energy of sensors is one promising solution to prolong their usage time. Many approaches have been developed to find the MC's moving path, and they assume that sensors have a fixed sensing rate (SR) and prefer to fully charge sensors. In practice, sensors can adaptively adjust their SRs to meet application demands or save energy. Besides, due to the fully charging policy, some sensors with low energy may take long to wait for the MC's service. Thus, the paper formulates a path and charge (P&C) problem, which asks how to dispatch the MC to visit sensors with adaptive SRs and decide their charging time, such that both survivability and throughput of sensors can be maximized. Then, we propose an efficient P&C scheduling (EPCS) algorithm, which builds the shortest path to visit each sensor. To make the MC fast move to charge the sensors near death, some sensors with enough energy are excluded from the path. Moreover, EPCS adopts a floating charging mechanism based on the ratio of workable sensors and their energy depletion. Simulation results verify that EPCS can significantly improve the survivability and throughput of sensors.

Service chaining is attracting attention as a promising technology for providing a variety of network services by applying virtual network functions (VNFs) that can be instantiated on commercial off-the-shelf servers. The data transmission for each service chain has to satisfy the quality of service (QoS) requirements in terms of the loss probability and transmission delay, and hence the amount of resources for each VNF is expected to be sufficient for satisfying the QoS. However, the increase in the amount of VNF resources results in a high cost for improving the QoS. To reduce the cost of utilizing a VNF, sharing VNF instances through multiple service chains is an effective approach. However, the number of packets arriving at the VNF instance is increased, resulting in a degradation of the QoS. It is therefore important to select VNF instances shared by multiple service chains and to determine the amount of resources for the selected VNFs. In this paper, we propose a cost-effective service chain construction with a VNF sharing model. In the proposed method, each VNF is modeled as an M/M/1/K queueing model to evaluate the relationship between the amount of resources and the loss probability. The proposed method determines the VNF sharing, the VNF placement, the amount of resources for each VNF, and the transmission route of each service chain. For the optimization problem, these are applied according to our proposed heuristic algorithm. We evaluate the performance of the proposed method through a simulation. From the numerical examples, we show the effectiveness of the proposed method under certain network topologies.

In this paper, we propose a selective membership inference attack method that determines whether certain data corresponding to a specific class are being used as training data for a machine learning model or not. By using the proposed method, membership or non-membership can be inferred by generating a decision model from the prediction of the inference models and training the confidence values for the data corresponding to the selected class. We used MNIST as an experimental dataset and Tensorflow as a machine learning library. Experimental results show that the proposed method has a 92.4% success rate with 5 inference models for data corresponding to a specific class.

Machine-to-Machine (M2M) communication plays a pivotal role in the evolution of Internet of Things (IoT). Cellular networks are considered to be a key enabler for M2M communications, which are originally designed mainly for Human-to-Human (H2H) communications. The introduction of M2M users will cause a series of problems to traditional H2H users, i.e., interference between various traffic. Resource allocation is an effective solution to these problems. In this paper, we consider a shared resource block (RB) and power allocation in an H2H/M2M coexistence scenario, where M2M users are subdivided into delay-tolerant and delay-sensitive types. We first model the RB-power allocation problem as maximization of capacity under Quality-of-Service (QoS) constraints of different types of traffic. Then, a learning framework is introduced, wherein a complex agent is built from simpler subagents, which provides the basis for distributed deployment scheme. Further, we proposed distributed Q-learning based autonomous RB-power allocation algorithm (DQ-ARPA), which enables the machine type network gateways (MTCG) as agents to learn the wireless environment and choose the RB-power autonomously to maximize M2M pairs' capacity while ensuring the QoS requirements of critical services. Simulation results indicates that with an appropriate reward design, our proposed scheme succeeds in reducing the impact of delay-tolerant machine type users on critical services in terms of SINR thresholds and outage ratios.

A neural conversational model (NCM) based on an encoder-decoder recurrent neural network (RNN) with an attention mechanism learns different sequence-to-sequence mappings from what neural machine translation (NMT) learns even when based on the same technique. In the NCM, we confirmed that target-word-to-source-word mappings captured by the attention mechanism are not as clear and stationary as those for NMT. Considering that vector norms indicate a magnitude of information in the processing, we analyzed the inner workings of an encoder-decoder GRU-based NCM focusing on the norms of word embedding vectors and hidden vectors. First, we conducted correlation analyses on the norms of word embedding vectors with frequencies in the training set and with conditional entropies of a bi-gram language model to understand what is correlated with the norms in the encoder and decoder. Second, we conducted correlation analyses on norms of change in the hidden vector of the recurrent layer with their input vectors for the encoder and decoder, respectively. These analyses were done to understand how the magnitude of information propagates through the network. The analytical results suggested that the norms of the word embedding vectors are associated with their semantic information in the encoder, while those are associated with the predictability as a language model in the decoder. The analytical results further revealed how the norms propagate through the recurrent layer in the encoder and decoder.

A surrogate-based electromagnetic (EM) optimization using neural networks (NNs) is presented for computationally efficient microwave bandpass filter (BPF) design. This paper first describes the forward problem (EM analysis) and the inverse problems (EM design), and the two fundamental issues in BPF designs. The first issue is that the EM analysis is a time-consuming task, and the second one is that EM design highly depends on the structural optimization performed with the help of EM analysis. To accelerate the optimization design, two surrogate models of forward and inverse models are introduced here, which are built with the NNs. As a result, the inverse model can instantaneously guess initial structural parameters with high accuracy by simply inputting synthesized coupling-matrix elements into the NN. Then, the forward model in conjunction with optimization algorithm enables designers to rapidly find optimal structural parameters from the initial ones. The effectiveness of the surrogate-based EM optimization is verified through the structural designs of a typical fifth-order microstrip BPF with multiple couplings.

Image quality assessment (IQA) is a fundamental metric for image processing tasks (e.g., compression). With full-reference IQAs, traditional IQAs, such as PSNR and SSIM, have been used. Recently, IQAs based on deep neural networks (deep IQAs), such as LPIPS and DISTS, have also been used. It is known that image scaling is inconsistent among deep IQAs, as some perform down-scaling as pre-processing, whereas others instead use the original image size. In this paper, we show that the image scale is an influential factor that affects deep IQA performance. We comprehensively evaluate four deep IQAs on the same five datasets, and the experimental results show that image scale significantly influences IQA performance. We found that the most appropriate image scale is often neither the default nor the original size, and the choice differs depending on the methods and datasets used. We visualized the stability and found that PieAPP is the most stable among the four deep IQAs.

In this paper, we propose new radio access network (RAN) architecture for reliable millimeter-wave (mmWave) communications, which has the flexibility to meet users' diverse and fluctuating requirements in terms of communication quality. This architecture is composed of multiple radio units (RUs) connected to a common distributed unit (DU) via fronthaul links to virtually enlarge its coverage. We further present grant-free non-orthogonal multiple access (GF-NOMA) for low-latency uplink communications with a massive number of users and robust coordinated multi-point (CoMP) transmission using blockage prediction for uplink/downlink communications with a high data rate and a guaranteed minimum data rate as the technical pillars of the proposed RAN. The numerical results indicate that our proposed architecture can meet completely different user requirements and realize a user-centric design of the RAN for beyond 5G/6G.

The phenomenon known as social polarization, in which a social group splits into two or more groups, can cause division of the society by causing the radicalization of opinions and the spread of misinformation, is particularly significant in online communities. To develop technologies to mitigate the effects of polarization in online social networks, it is necessary to understand the mechanism driving its occurrence. There are some models of social polarization in which network structure and users' opinions change, based on the quantified opinions held by the users of online social networks. However, they are based on the interaction between users connected by online social networks. Current recommendation systems offer information from unknown users who are deemed to have similar interests. We can interpret this situation as being yielded non-local effects brought on by the network system, it is not based on local interactions between users. In this paper, based on the spectral graph theory, which can describe non-local effects in online social networks mathematically, we propose a model of polarization that user behavior and network structure change while influencing each other including non-local effects. We investigate the characteristics of the proposed model. Simultaneously, we propose an index to evaluate the degree of network polarization quantitatively, which is needed for our investigations.

Recently, multi-modal fusion methods based on remote sensing data and social sensing data have been widely used in the field of urban region function recognition. However, due to the high complexity of noise problem, most of the existing methods are not robust enough when applied in real-world scenes, which seriously affect their application value in urban planning and management. In addition, how to extract valuable periodic feature from social sensing data still needs to be further study. To this end, we propose a multi-modal fusion network guided by feature co-occurrence for urban region function recognition, which leverages the co-occurrence relationship between multi-modal features to identify abnormal noise feature, so as to guide the fusion network to suppress noise feature and focus on clean feature. Furthermore, we employ a graph convolutional network that incorporates node weighting layer and interactive update layer to effectively extract valuable periodic feature from social sensing data. Lastly, experimental results on public available datasets indicate that our proposed method yeilds promising improvements of both accuracy and robustness over several state-of-the-art methods.

Stance prediction on social media aims to infer the stances of users towards a specific topic or event, which are not expressed explicitly. It is of great significance for public opinion analysis to extract and determine users' stances using user-generated content on social media. Existing research makes use of various signals, ranging from text content to online network connections of users on these platforms. However, it lacks joint modeling of the heterogeneous information for stance prediction. In this paper, we propose a self-supervised heterogeneous graph contrastive learning framework for stance prediction in online debate forums. Firstly, we perform data augmentation on the original heterogeneous information network to generate an augmented view. The original view and augmented view are learned from a meta-path based graph encoder respectively. Then, the contrastive learning among the two views is conducted to obtain high-quality representations of users and issues. Finally, the stance prediction is accomplished by matrix factorization between users and issues. The experimental results on an online debate forum dataset show that our model outperforms other competitive baseline methods significantly.

Cyber-physical systems (CPSs) assisted by digital twins (DTs) integrate sensing-actuation loops over communication networks in various infrastructure services and applications. This study overviews the concept, methodology, and applications of the integrated communication quality estimation and control for the DT-assisted CPSs from both communications and control perspectives. The DT-assisted CPSs can be considered as networked control systems (NCSs) with virtual dynamic models of physical entities. A communication quality estimation observer (CQEO), which is an extended version of the communication disturbance observer (CDOB) utilized for time-delay compensation in NCSs, is proposed to estimate the integrated effects of the quality of services (QoS) and cyberattacks on the NCS applications. A path diversity technique with the CQEO is also proposed to achieve reliable NCSs. The proposed technique is applied to two kinds of NCSs: remote motor control and haptic communication systems. Moreover, results of the simulation on a haptic communication system show the effectiveness of the proposed approach. In the end, future research directions of the CQEO-based scheme are presented.

In recent years, deep neural networks (DNNs) have made a significant impact on a variety of research fields and applications. One drawback of DNNs is that it requires a huge amount of dataset for training. Since it is very expensive to ask experts to label the data, many non-expert data collection methods such as web crawling have been proposed. However, dataset created by non-experts often contain corrupted labels, and DNNs trained on such dataset are unreliable. Since DNNs have an enormous number of parameters, it tends to overfit to noisy labels, resulting in poor generalization performance. This problem is called Learning with Noisy labels (LNL). Recent studies showed that DNNs are robust to the noisy labels in the early stage of learning before over-fitting to noisy labels because DNNs learn the simple patterns first. Therefore DNNs tend to output true labels for samples with noisy labels in the early stage of learning, and the number of false predictions for samples with noisy labels is higher than for samples with clean labels. Based on these observations, we propose a new sample selection approach for LNL using the number of false predictions. Our method periodically collects the records of false predictions during training, and select samples with a low number of false predictions from the recent records. Then our method iteratively performs sample selection and training a DNNs model using the updated dataset. Since the model is trained with more clean samples and records more accurate false predictions for sample selection, the generalization performance of the model gradually increases. We evaluated our method on two benchmark datasets, CIFAR-10 and CIFAR-100 with synthetically generated noisy labels, and the obtained results which are better than or comparative to the-state-of-the-art approaches.

Real-time applications are becoming more and more popular, and due to the demand for more compact and portable user devices, offloading terminal processes to edge servers is being considered. Moreover, it is necessary to process packets with low latency on edge servers, which are often virtualized for operability. When trying to achieve low-latency networking, the increase in server power consumption due to performance tuning and busy polling for fast packet receiving becomes a problem. Thus, we design and implement a low-latency and energy-efficient networking system, energy-efficient kernel busy poll (EE-KBP), which meets four requirements: (A) low latency in the order of microseconds for packet forwarding in a virtual server, (B) lower power consumption than existing solutions, (C) no need for application modification, and (D) no need for software redevelopment with each kernel security update. EE-KBP sets a polling thread in a Linux kernel that receives packets with low latency in polling mode while packets are arriving, and when no packets are arriving, it sleeps and lowers the CPU operating frequency. Evaluations indicate that EE-KBP achieves microsecond-order low-latency networking under most traffic conditions, and 1.4× to 3.1× higher throughput with lower power consumption than NAPI used in a Linux kernel.

The recurrent neural network (RNN) has been used in audio and speech processing, such as language translation and speech recognition. Although RNN-based architecture can be applied to speech synthesis, the long computing time is still the primary concern. This research proposes a fast gated recurrent neural network, a fast RNN-based architecture, for speech synthesis based on the minimal gated unit (MGU). Our architecture removes the unit state history from some equations in MGU. Our MGU-based architecture is about twice faster, with equally good sound quality than the other MGU-based architectures.