This paper proposes a novel design of an intelligent, personalized travel recommendation service (TRS), and presents a prototypical implementation with practical field tests in Penghu islands, Taiwan. The proposed TRS can import historical tourist footprints, personal preferences, real-time environmental data like weather and crowd conditions. It then applies big data analytics and XGBoost-based machine learning techniques on Google cloud-based microservice platform to generate personalized travel itineraries against dynamic local travel conditions. Practical demonstration illustrates enhanced user experience by intelligent recommendation of tourist attractions.

Owing to the widespread application of machine learning, increasing attention has been focused on extensive data collection for learning model construction. Recently, with growing concerns about data privacy, private information protection has significantly increased the operation cost and difficulty of boosting model performance. The Federated Learning (FL) technique has been introduced to address this issue by keeping data on client devices and reducing the need to handle sensitive data directly. However, several challenging issues may arise when applying FL, such as data heterogeneity, efficient feature transmission, and additional computational demands. In this study, a novel FL model, Elastic-Trust Hybrid Federated Learning (ET-FL), is introduced with a dual federated learning framework. ET-FL incorporates the trust mechanism and differential aggregation strategy for model optimization and computation reduction. In addition, the proposed model is applied on real-world datasets to show the performance and practicability of promising results.

Mobile crowdsensing (MCS) technologies usher in new distributed services that utilize user-provided resources to execute tasks in ubiquitous network environments. Conventional studies mainly emphasize user recruitment for simple task allocation with neither information exchange nor collaboration between users. Our study aims to investigate the new issue of collaborative multi-user (CMU) task allocation, while many users cooperate in complex tasks that require multi-fold resources from different users. To form a group of collaborative users, we adopt the social-tie notion to represent three group types, including basic, bridging, and linked user groups, in MCS contexts. By quantifying the strength of social connections between users, we formulate the measures of service capacity and cost with respect to any user group. Then, the fittingness of each group to a task can be calculated, and then a subset of user groups can be chosen to perform CMU tasks. We propose a group-based task allocation scheme, briefly named GTA, which can evaluate the cost-effectiveness of task processing by any particular group and thus allocate appropriate groups in accordance with different task requirements. Performance results by simulation manifest that the GTA scheme is promising in sustaining lower service time with only a minor influence on service cost, as compared with two typical schemes, MinCost and GoCC.

In recent years, social networks have grown in popularity, with most people actively engaging on these platforms. These networks hold valuable insights into users' values and interests, allowing us to analyse relationships between connected individuals and even predict potential friendships. However, social networks are dynamic, and their structure evolves over time. To account for this, we employed a dual approach using a bi-phase LSTM autoencoder and a bi-phase LSTM predictor. These tools capture the changing characteristics of social networks and predict future graph structures. We rigorously tested our model on three datasets and compared its performance with other models. The bi-phase LSTM consistently delivered strong results across all datasets. Additionally, the model's hyperparameters were fine-tuned to improve predictive accuracy, demonstrating its reliability in forecasting the evolution of social network structures.

Matrix factorization (MF) technique has been widely utilized in recommendation systems due to the precise prediction of users’ interests. Prior MF-based methods adapt the overall rating to make the recommendation by extracting latent factors from users and items. However, in real applications, people’s preferences usually vary with time; the traditional MF-based methods could not properly capture the change of users’ interests. In this paper, by incorporating the recurrent neural network (RNN) into MF, we developed a novel recommendation system, M-RNN-F, to effectively describe the preference evolution of users over time. A learning model is proposed to capture the evolution pattern and predict the user preference in the future. The experimental results show that M-RNN-F performs better than other state-of-the-art recommendation algorithms. In addition, we conduct experiments on real world dataset to demonstrate the practicability.

Consider the high dynamics of traffic loading and resource provision on network hosts that forward data flows along a particular path between two endpoints. The Quick UDP Internet Connect (QUIC) protocol performs better than TCP for its effects in shortening the time of connection establishment and data transmission between two endpoints. Recent studies attempted to exploit the notion of multipath QUIC that forwards the data over multiple paths. Using the multipath QUIC can not only augment the total bandwidth capacity but also avoid traffic congestion on some paths. In this paper, our study proposes a novel multipath QUIC scheme which is able to minimize the flow completion time of multipath QUIC by jointly utilizing two measures of path delay and packet loss rate on a path. Experimental results show that the proposed algorithm is superior to other scheduling schemes, including naive QUIC and Lowest-RTT-First QUIC.

Data about vehicle trajectories assumes a crucial role in applications such as intelligent connected vehicles. However, missing values resulting from sensors and other factors frequently affect real trajectory data. Currently, it is challenging to utilize trajectory completion methods to generate accurate real-time results at an affordable computing cost. This paper proposes GNN-RM, a trajectory completion algorithm based on graph neural networks and regeneration modules, encompassing feature extraction, subgraph construction, spatial interaction graph, and trajectory regeneration modules. The feature extraction algorithm extracts influential data as feature vectors based on certain conditions and organizes these feature vectors into different subgraphs according to categories. The spatial interaction graph constructed through graph neural networks extracts spatial interaction features between vehicles and the environment, while the regeneration modules constructed by multi-head attention mechanisms extract temporal features of vehicles, thereby completing the missing trajectories. The experimental results demonstrate that GNN-RM can achieve higher trajectory completion accuracy with fewer input parameters than multiple baseline models.

Recently, the importance of few-shot learning has tremendously grown due to its widespread applicability. Via few-shot learning, users can train their models with few data and maintain high generalisation ability. Meta-learning and continual learning models have demonstrated elegant performance in model development. However, unstable performance and catastrophic forgetting are still two fatal issues with regard to retaining the memory of knowledge about previous tasks when facing new tasks. In this paper, a novel method, enhanced model-agnostic meta-learning (EN-MAML), is proposed for blending the flexible adaptation characteristics of meta-learning and the stable performance of continual learning to tackle the above problems. Based on the proposed learning method, users can efficiently and effectively train the model in a stable manner with few data. Experiments show that when following the N-way K-shot experimental protocol, EN-MAML has higher accuracy, more stable performance and faster convergence than other state-of-the-art models on several real datasets.

The rapid growth of network services and applications has led to an exponential increase in data flows on the internet. Given the dynamic nature of data traffic in the realm of internet content distribution, traditional TCP/IP network systems often struggle to guarantee reliable network resource utilization and management. The recent advancement of the Quick UDP Internet Connect (QUIC) protocol equips media transfer applications with essential features, including structured flowcontrolled streams, quick connection establishment, and seamless network path migration. These features are vital for ensuring the efficiency and reliability of network performance and resource utilization, especially when network hosts transmit data flows over end-to-end paths between two endpoints. QUIC greatly improves media transfer performance by reducing both connection setup time and transmission latency. However, it is still constrained by the limitations of single-path bandwidth capacity and its variability. To address this inherent limitation, recent research has delved into the concept of multipath QUIC, which utilizes multiple network paths to transmit data flows concurrently. The benefits of multipath QUIC are twofold: it boosts the overall bandwidth capacity and mitigates flow congestion issues that might plague individual paths. However, many previous studies have depended on basic scheduling policies, like round-robin or shortest-time-first, to distribute data transmission across multiple paths. These policies often overlook the subtle characteristics of network paths, leading to increased link congestion and transmission costs. In this paper, we introduce a novel multipath QUIC strategy aimed at minimizing flow completion time while taking into account both path delay and packet loss rate. Experimental results demonstrate the superiority of our proposed method compared to standard QUIC, Lowest-RTT-First (LRF) QUIC, and Pluginized QUIC schemes. The relative performance underscores the efficacy of our design in achieving efficient and reliable data transfer in real-world scenarios using the Mininet simulator.

Deploying flying ferries in large-scaled wireless sensor networks can prevent data gathering and distribution from physical communication restrictions on the ground. Ferry-assisted data distribution is straightfoward and manageable, as compared with traditional ad hoc routing over weak and multi-hop wireless networks. Our study exploits the notion of a ferry fleet that conducts teamwork for data gathering in such environments. To maximize the lifetime of a ferry fleet, we propose a fair and energy-balanced ferry fleet placement scheme, named as FEB for brevity, as jointly considering the efficiency, balance, and fairness of energy consumption. This scheme operates two mutual phases. The first phase demarcates service regions in a network using the powered-Voronoi diagram. The second phase decides a shortest-path itinerary across sensors in every service region using a genetic-based alternative of the traveling salesman problem (TSP) algorithm. Thus, this scheme is able to ensure fair task assignment, balance energy consumption, and prolong the lifetime among multiple ferries in teamwork. Performance results show that the proposed scheme outperforms several typical schemes, including Native, K-Means and Spiral, in terms of cumulative energy computation, residual energy distribution, Jain's fairness index on energy utilization, the number of alive ferries, and the total of task execution times during ferry teamwork.

The population of the Chinese white dolphin is claimed to be critically endangered and is on the International Union for Conservation of Nature (IUCN) Red list. It is estimated that there are fewer than 100 individuals in the East Taiwan Strait, and the number is falling. The dolphin’s habitat has been seriously impacted by man-made pollution, such as industry contamination, fishing, and noise. To prevent extinction of the species, conservative action is vital. Prior to any such action, data on the dolphin are essential for decision makers. The current method of observing dolphins is the man-on-boat-watch approach, which is heavily dependent on manpower. Its performance is seriously affected by the weather, fatigue of those on board, and it is also risky and costly. An Internet of things (IoT) data collection mechanism concept is proposed for the purpose of observing dolphins with close watch. It consists of off-the-shelf products such as hydrophones, unmanned aerial vehicles (UAVs) and a specific command and control in search/detection for carrying out the observation task. A Monte Carlo simulation model was developed to analyze the effectiveness of the feasible alternatives, in which some factors are considered and analyzed for their significance. The simulation result showed that the IoT mechanism has an 8.5 times greater chance of availability in operation and at least 2 times more contact than the man-on-boat-watch method. The significant factor affecting the IoT mechanism’s effectiveness is the number of hydrophones and UAVs in the scenario. The great contribution made by this study is that it is the first analytical paper to reveal the effectiveness of an IoT mechanism in benefitting the observation of Chinese white dolphin. The limitation is that it uses off-the shelf products for the IoT mechanism instead of high-end products which could be more effective.

The Chinese white dolphin, along the west coast of Taiwan, is claimed as on the brink of distinction species. A top priority before entering the policy process of conservation is to have a well plan of accurate observation. Objective of this paper is to evaluate a proposed novel alternative, which applies the internet of thing (IoT) concept, in the observation of wild animal instead of the conventional way which is manpower intensive. The IoT observation system is consisted of the hydrophone laid on the sea floor, camera/IR sensor carried by quadcopter, and land control center. Monte Carlo simulation was developed for getting insight of the interaction between dolphin and IoT observation system. The result showed IoT observation system has a significant effect. The contribution this paper made is to prove that the IoT observation system is far more cost effectively than the manpower intensive boat watching method.

With the emerging Internet of Things technology, the world is facing rapid changes in all areas; firefighting is no exception. Conventional firefighting is a dangerous occupation which involves saving lives and property from fires. The skills of firefighting have not changed greatly over the years; hence, using the IoT to aid firefighters is a way to improve their performance. Due to the lack of research on implementing the IoT in the firefighting domain, the objective of this study was to use the quantitative method to gain insights into the usefulness of using the IoT as an aid to firefighting. A Monte Carlo simulation was developed for processing the detailed firefighting interactions in situations of uncertainty. After the verification of the simulation model, the results showed that the search time ratios of unmanned aerial vehicle (UAV) to conventional firefighting for various levels of severity of fire were 30.09, 26.69, and 22.24%. The search and rescue time ratios of UAV to conventional firefighting were 48.27, 35.95, and 31.87%. The most important of these statistics is that at least 50% of the time spent by firefighters on the scene of the fire can be reduced by using the Internet of Things. All of the above data were analyzed usingt test, which showed significant improvement when the Internet of Things was implemented in firefighting. The contribution of this study is to present quantitative results for proving the value of integrating the Internet of Things into firefighting.

Real-time finger detection and tracking systems have been growing rapidly in the past decade. Among those methods, Appearance-based and Model-based methods have produced excellent results. However, the occlusion issue is one of the main challenges in this field. In this study, we address this issue by considering the repeating-finite gestures of a guitar-strumming or a hand puppet and, represent using a Finite State Machine model. Also we proposed a novel finger pose tracking system using FSM Model combining with the appearance-based method.. The proposed system consists of two parts: FSM-FT builder creates the FSM hand, and the FSM-FT runner controls the FSM-FT system. Empirically, we conducted an experimental study involving one sample repeating hand gesture and our approach achieved a significance recognition rate of 82% in the testing phase.

With the rapid development and massive deployment of the Internet of things (IoT) networks, security related issues in the IoT networks have been paid more and more attention to. Among all the security concerns, message authentication is critical in preventing the unauthorized messages from being transmitted in the IoT networks. Many message authentication schemes have been proposed based on the public-key cryptosystem, where the key management is simple and scalable. Identity based cryptosystem is a special type of public-key cryptosystem and can further ease the process of the key management since the public keys can be obtained easily. In this paper, we devise an efficient message authentication with enhanced privacy (IMAEP) scheme using the identity based signature. Our proposed scheme can provide both unconditional privacy as well as the enhanced privacy under full key exposure attack. Our proposed scheme can also provide existential unforgeability under the adaptive chosen-message-and-identity attack. Compared with the scheme that has the same level of anonymity and security, our proposed scheme has much lower computational overhead, and can provide extra unconditional privacy. Next we propose an extended IMAEP (EIMAEP) scheme for the general access structures where the message is signed by a group of users instead of one user. We also conduct comprehensive analysis and demonstrate that the EIMAEP scheme can achieve the same level of privacy and unforgeability as the IMAEP scheme.

Routing in delay-tolerant networks exploits node mobility to distribute messages among inter-contact of mobile nodes in intermittently connected environments. However, only depending on node movements to deliver messages between any two nodes in a network could result in lower successful delivery ratio, longer latency and high transmission overhead. Recent research intends to add ferries which can itinerate around some strategic places in a network to expedite message distribution and improve network performance. This paper designs a ferry travel scheduling approach for efficiently message forwarding in delay-tolerant networks. In this approach, multiple ferries attempt to sense the intensity of message exchanges in geographic proximity, discover hotspot areas by contact history and then autonomously plan travel routes to go through those hot spots. To examine the network performance elevated by ferries, this paper incorporates this ferry-aided approach into several typical delay-tolerant routing schemes, including Epidemic, Spray and Wait, and PRoPHET. Relative comparison among these combined schemes is shown under synthetical Random Waypoint and real NCUTrace mobility models.

There are many animal species have extinct from the earth because of the behavior of human being, such as killing and the change of habitat environment due to climate change, contamination and pollution, etc. The close-in observation is a way for counting and identifying the animal, which is on the verge of distinction. The conventional way in animal observation is by vehicle/foot on land and by vessel at sea that is not only man-power consuming, but costly, ineffective and weather limited. This paper aims at using a new information technology for dolphin observation instead of man-power on board vessel. A concept of responsive action in observation operation is applied carrying out by an information-based observation system which consists of hydrophone laid on the sea floor, camera/IR camera carried by Quadrocopter, and land control center. For verifying the effectiveness of this information-based observation system, the Monte Carlo simulation model is developed. The results showed this information-based observation system is more effective than the man-power, and the concept of the search in area of uncertainty (AOU) can make a significant improvement in detecting dolphin.

The internet of things (IoT) has become a trend in interactive environments for providing information to decision-makers. Anti-submarine warfare (ASW) is a typical pursuit and evasion (PE) game that is a very complicated process. The ASW helicopter is assigned to execute the final phase of hunting the submarine with a torpedo attack. In most cases, a single helicopter is assigned to detect the submarine by dipping sonar, and then drops a torpedo. Once the dipping sonar goes off, uncertainty takes over, with the possible result of losing track of the submarine. To prevent this problem, using the IoT concept to create a wireless sensor network (WSN) in the area of interest for keeping ears on the evading submarine is a potential solution. The objective of this paper is to gain insights into this PE scenario so as to quantify the interaction result in order to demonstrate the effectiveness of the helicopter in terms of hunting the submarine. Monte Carlo simulation has been developed as the analytical tool, and ANOVA was used to verify the significance of the output measure of effectiveness (MOE) before analysis. The results show that a slow, unalerted submarine has a very low chance of survival. An alerted submarine has very high chance of survival, but when the proposed sonobuoy WSN is in place, this situation benefitting the submarine will be reversed. The WSN has been proved to be effective in a single helicopter carrying out its ASW task.

Nowadays, fire accident is still a thorny problem due to the current firefighting still heavily relies on the experience instead of information. Saving lives from fireground is the primary task in firefighting, in which the speed of effective search largely relies on the sufficient and instant information. When insufficient information situation follows firefighter tightly, the firefighter's life can be jeopardised. Equipped firefighter with the advancing information technology, such as IR, laser range-finder, camera, augmented reality and an unmanned aerial vehicle for acquiring more fireground information may be useful for firefighting task. This study focuses on using Monte Carlo simulation model to quantify the feasible alternatives and finds out the significant effect by t-test. The result showed the time in search of victim is significantly reduced as using new way of firefighting. The contribution of this paper is to disclose the value of the proposed information-based technology in support of firefighting.