Location-Based Social Networks (LBSNs) offer a rich dataset of user activity at Points-of-Interest (POIs), making next POI recommendation a key task. Traditional algorithms face challenges due to broad searching scopes, affecting recommendation accuracy. Users tend to visit nearby POIs and show temporal concentration in their activities, reflecting personalized spatio-temporal clustering. However, individual user data may be insufficient to capture these clustering effects for personalized recommendations. In this paper, we propose an integrated Personalized Spatio-Temporal Clustering Model (iPCM) for next POI recommendation. The model learns this kind of personalized spatio-temporal clustering effect by using global historical trajectory data in conjunction with user feature embeddings. It integrates the features of personalized spatio-temporal clustering with the user's trajectory, and completes the user's POI recommendation through a Transformer encoding and MLP decoding. To enhance the accuracy of predictions, we add a module of probability adjustment. The experimental results on multiple datasets show that with the help of personalized spatio-temporal clustering, the proposed iPCM is superior to existing methods in various evaluation metrics.

Knowledge Graph based Retrieval-Augmented Generation (KG-RAG) technology is an effective method to improve the performance of Large Language Models (LLMs) in medical question answering tasks. However, there is an inherent difference between structured knowledge graphs and sequential LLMs. The existing method of converting triples of knowledge graph into text through predefined templates leads to loss of structural information and excessive context length, resulting in alignment difficulties between the two. In this paper, we propose a GNN-LLM Aligned Medical knowledge QuestionAnswering model (GLAM-QA). The model first pretrains a GNN to extract structural information from the knowledge graph, realizes the interactive fusion of graph representations and LLM vocabulary embeddings through a cross attention layer, and then generates graph tokens via a projector for input to the LLM. It can retain graph structural information without finetuning the LLM and provide accurate and concise answers. Experiments on three public datasets show that compared with five comparative algorithms, the proposed GLAM-QA achieves better performance in BertScore, F1-Score, and BLEU-1 metrics. This research provides an effective solution to the alignment problem between knowledge graphs and LLMs, and improves the accuracy and practicality of medical question answering systems.

With the rapid development of high-throughput sequencing technologies, multi-omics integration analysis has become a core means to decipher complex disease mechanisms, and graph neural networks (GNNs) have shown significant advantages in multi-omics data fusion due to their strong nonlinear modeling and relational reasoning capabilities. However, such models typically rely on large-scale labeled data for training, while small-sample disease scenarios are prevalent in biomedical research. The limited sample size in these scenarios does not support stable deep model training, severely restricting the application of precision medicine in critical disease fields. Directly constructing GNN models under small-sample settings leads to severe overfitting and training instability, resulting in a significant decline in model generalization performance. In this paper, we propose a Cross-disease knowledge transfer framework for Small-sample Omics analysis with GNNs (CSOG), through a pretraining and parameter-freezing fine-tuning paradigm. To the best of our knowledge, we are the first to explore the GNN cross-disease knowledge transfer framework for small-sample omics analysis. Experimental evaluations on three independent small-sample datasets show that compared with 10 state-of-theart baseline methods, the proposed method achieves significantly improved average classification accuracy and maintains stability across different sample sizes.

Graph computing in cloud-edge collaborative environments faces critical challenges in distributed task processing, particularly in fundamental operations such as subgraph isomorphism that underpins triangle counting applications. In typical architectures where data streams are transmitted from edge collectors to cloud masters, conventional approaches employ reservoir sampling to distribute edge streams among workers for triangle estimation. However, the computational accuracy degradation is caused by cross-domain edge distribution strategies. In this paper, we propose a cloud-edge collaborative framework for distributed triangle counting. We employ spectral clustering analysis to reveal latent domain relationships that guide edges distribution. Our experimental evaluation uses streaming data with global relative error measurement across multiple datasets, demonstrating superior performance over existing algorithms.

In streaming federated learning, where data on each client is received in the form of a data stream, the distribution of data on the clients has a significant impact on the performance of the federated learning model. The continuous influx of streaming data on the clients leads to real-time changes in the local data distribution, which in turn affects the performance of the federated learning model. Furthermore, the heterogeneity in data distribution among clients exacerbates this impact. In this paper, to address these challenges, we propose a Data-aware Probabilistic Client Sampling scheme (DPCS) for selecting appropriate clients to participate in model training in each round of federated learning. DPCS begins with a method for real-time monitoring of local data distributions on the clients. Based on these observations, the central server adopts a probability-based client sampling strategy. Through extensive experimentation, we demonstrate that our client sampling scheme offers higher timeliness and enhances the performance of federated learning compared to traditional methods.

Streaming triangle counting is a critical issue in graph stream mining, with applications in dense subgraph discovery, web mining, anomaly detection, and more. Recent efforts have focused on estimating triangle counts in graph streams, primarily through sampling methods. However, because of limited memory resources for handling high speed streams, traditional sampling methods suffer from reduced sampling rate and thereby performance loss. In this paper, we propose a new compact data structure called uHLL to process edge streams by considering the tradeoff between estimation accuracy and memory efficiency. Furthermore, different from conventional triangle counting algorithms, we solve the estimation of union set cardinality for edge-local triangle count under both centralized and distributed framework, so as to efficiently estimate the global triangle count by a one-pass streaming algorithm. To the best of our knowledge, this is the first implementation of a distributed framework using a compact data structure for streaming triangle counting. We provide theoretical proof of unbiasedness and derive the variance of the union set and global triangle count. We compare our scheme with 11 algorithms, showing that under the same experimental setting, uHLL and distributed uHLL are at least$ 2.3$and$ 1.7$times more accurate than the state-of-the-art, respectively.

RFID (Radio Frequency Identification) computing devices in practical applications often suffer from their poor computing performance in terms of low task throughput (also known as task rate) due to scarce harvested power. For optimizing the task throughput, the basic idea is to choose an optimal processor clock frequency when executing a specific task ($e.g.$, operate sensor) in order to maximize task execution rate. Existing methods are based on the common sense where the frequency and task throughput are directly proportional to each other, meaning that a higher frequency causes a higher task rate. In RFID-scale devices, however, we observe that the relationship between the frequency and task throughput overturns the common sense, in which if the device rises the frequency, the task throughput will increase first and then decrease due to intermittent task execution pattern on such devices. In this paper, we present a systematic task throughput model to explain and formulate the non-monotonic relationship between the frequency and task throughput. Based on the throughput model, we further introduce dynamic optimal frequency scaling (DOFS) to calculate the optimal frequency for task execution and thus optimize the task throughput in the RFID-scale devices. The experimental results show that the task throughput can be improved by 45.8% on average compared to the existing best effort.

Recently, the counting algorithm of local topology structures, such as triangles, has been widely used in social network analysis, recommendation systems, user portraits and other fields. At present, the problem of counting global and local triangles in a graph stream has been widely studied, and numerous triangle counting steaming algorithms have emerged. To improve the throughput and scalability of streaming algorithms, many researches of distributed streaming algorithms on multiple machines are studied. In this article, we first propose a framework of distributed streaming algorithm based on the Master-Worker-Aggregator architecture. The two core parts of this framework are an edge distribution strategy, which plays a key role to affect the performance, including the communication overhead and workload balance, and aggregation method, which is critical to obtain the unbiased estimations of the global and local triangle counts in a graph stream. Then, we extend the state-of-the-art centralized algorithm TRIÈST into four distributed algorithms under our framework. Compared to their competitors, experimental results show that DVHT-i is excellent in accuracy and speed, performing better than the best existing distributed streaming algorithm. DEHT-b is the fastest algorithm and has the least communication overhead. What’s more, it almost achieves absolute workload balance.

Differing from tremendous existing works that mainly focus on optimizing backscatter communication, Radio-to-Bus (R2B) communication utilizes backscatter to offload processors from IoT devices to the gateway, achieving processor-free devices of significantly reduced power and hardware cost. However, R2B communication is not suitable for large-scale backscatter networks, since R2B cannot support parallel and long-range communication between the gateway and hundreds of R2B devices. In this article, we present Chipnet, a network that supports hundreds of long-range and concurrent connections between the gateway and multiple processor-free devices. The high-level design of Chipnet includes a parallel frequency-division uplink mechanism that can work on processor-free devices and a processor-free MAC layer protocol that supports gateway to broadcast downlink data and individually manage each processor-free device. This design addresses practical issues facing the processor-free device architecture, such as synchronizing hundreds of processor-free devices, assigning unique channel frequencies to every device, and realizing power-efficient processor-free signal conversion. The results demonstrate that a Chipnet network can achieve a task throughput of 2,400 tasks/s with a latency of 72.23 ms. Compared with the R2B network, Chipnet achieves 3×–5× improvements in network coverage range and two orders of magnitude improvement in both network throughput and network latency.

Mobile Edge Caching (MEC) can cache content at the edge of the network to reduce the delay and overhead of content transmission, which has become an effective method to solve the explosive growth of network traffic. To make good use of the limited resources in edge devices, many contents caching strategies use various methods to predict the popularity of content. However, caches get close to the edge of the network can lead to the rapid increase of caches' number and the user's requests are dispersed into a large number of caches, which leads to the popularity distribution of contents in edge caches is quite different and the number of first miss requests (the corresponding content is requested for the first time and is not in the cache) in edge caches becoming an essential factor affecting the cache hit rate. This paper first demonstrates the significant impact of the first miss requests through dataset analysis and establishes a mathematical model for the first miss problem in the edge cache. Then we analyze the similarity of requests received by caches and propose a proactive push algorithm based on similarity to improve the hit rate of edge caches. Through the trace-driven simulation experiment, we verify that the methods proposed in this paper can significantly improve the caches' hit rate.

Personalized trip recommendation tries to recommend a sequence of point of interests (POIs) for a user. Most of existing studies search POIs only according to the popularity of POIs themselves. In fact, the routes among the POIs also have attractions to visitors, and some of these routes have high popularity. We term this kind of route as Attractive Route (AR), which brings extra user experience. In this paper, we study the attractive routes to improve personalized trip recommendation. To deal with the challenges of discovery and evaluation of ARs, we propose a personalized Trip Recommender with POIs and Attractive Route (TRAR). It discovers the attractive routes based on the popularity and the Gini coefficient of POIs, then it utilizes a gravity model in a category space to estimate the rating scores and preferences of the attractive routes. Based on that, TRAR recommends a trip with ARs to maximize user experience and leverage the tradeoff between the time cost and the user experience. The experimental results show the superiority of TRAR compared with other state-of-the-art methods.

Recently, the counting algorithm of local topology structures, such as triangles, has been widely used in social network analysis, recommendation systems, user portraits and other fields. At present, one-pass streaming algorithm for counting global and local triangles has been widely studied, and most researches focus on the single-machine streaming algorithm in a 'offline+batch processing' mode. However, researches on distributed online algorithm on multiple machines are still in its infancy, and this stage has not been thoroughly studied. In this paper, we investigate the triangle counting problem in large-scale simple undirected graphs whose edges arrive as a stream. We propose two distributed online streaming algorithms to estimate the global number of triangles, which are based on the current best performance sampling-based streaming algorithm. We mainly realize the reasonable partition of the graph stream, so that each worker independently estimates the number of triangles in a subgraph of the graph stream. Experimental results show that our algorithms reduce the estimation error and are several times more accurate than state-of-the-art streaming algorithms.

Pedestrian flow is an important indicator of public places, since it can provide more potential economic benefits. Pedestrian flow prediction is developed to help the decisionmaking for the operators (such as shopping center owner). Furthermore, the operators aperiodically arrange some events to attract the nearby pedestrians, such as the sales promotions, and we term this kind of events as business event. Moreover, their placement will affect the distributions of the pedestrian flows. In this paper, we investigate the influence of the business events on the pedestrian flows. Then, we propose an Attraction Based Matrix Factorization model, called ABMF, to efficiently predict the pedestrian flow with business events and enable operators to formulate candidate solutions. The experimental results show the superiority of our prediction method compared with other state-of-the-art prediction techniques.

To improve the experience of various network applications, dynamic rate allocation is an essential issue in recent ISP networks. The emergence of software-defined networking (SDN) and the OpenFlow specification makes dynamic rate allocation in ISP networks efficient. The allocation could locally run on a home network gateway (edge switch) under SDN, but such local range of rate allocation reduces the overall fairness and performance in the whole network. However, under a global range, the request of rate allocation from a small number of hosts will cause all switches on the entire network to participate. This is termed as cascading problem, which causes a high cost for re-allocating the rates with the global range of switches. In this paper, we investigate the cascading problem for dynamic rate allocation with SDN, and discuss the tradeoff between the performance and cost for the range of rate allocation. We propose a Rate Allocation algorithm with Limited Range (RALR) in SDN, and discuss it for dynamic rate allocation by the theory of Lyapunov drift. Our intensive simulations verify the performance of the proposed strategy of rate allocation in SDN.

Software Defined Networks (SDN) have been proposed for use in applications of the Internet of Things (IoT), termed as software-defined IoT (SD-IoT) network, because of the popularity and capability of mobile devices being used for networking in relatively large areas. However, a single controller in SDN has a limited request-processing capability, so a distributed control plane with multiple physical controllers has been used to achieve scalability and reliability for supporting the IoT applications. Accordingly, the data plane of an SDN is partitioned into multiple domains, and each controller just takes over one. When considering both delays and loads of requests to the controllers, a partitioning problem arises. It is required to consider the distributions of flow paths, since inter-domain flow paths will create an extra load of requests to the controllers. In this paper, we investigate the partitioning problem in SD-IoT networks. Since uploading sensing data through the IoT gateways are non-uniform, we utilize a hypergraph to model the relationship between the spatial events and the gateways in IoT for urban sensing. We propose a Partitioning Algorithm for Software-defined IoT Network (PASIN) to partition the SDN by considering both delays and loads of requests from the flow paths. Our extensional simulations verify the effectiveness of our proposed approach.

Software Defined Networks (SDN) have been utilized in applications of the Internet of Things (IoT), termed as software-defined IoT network, because of the popularity and capability of mobile devices being used for networking in relatively large areas. In a software-defined IoT system, the sensing data are asynchronously harvested by the mobile sensing nodes, and are also asynchronously uploaded to the gateways of a software-defined network. Thus, all the sensing data are asynchronously transmitted from the gateways to the data servers in the pattern of multipoints-to-point (M2P) data transmissions. Even if the sensing data are generated from the same sensing event, the controller of SDN has no knowledge about this relationship. Thus, such asynchronous M2P data transmissions from the same sensing event at the gateways will generate many redundant requests to their controller by OpenFlow protocol of SDN. In this paper, we investigate the redundant requests caused by the asynchronous M2P data transmissions in the software-defined IoT network. We model the relationship between the sensing events and the uploading gateways by utilizing their spatial locations and the distribution of mobile sensor nodes. To reduce the loads on the controller for the asynchronous M2P data transmissions, we propose an One-Request Scheme for Software-Defined IoT Networks (ORSIN), to batch the updating the forwarding rules of the multiple data transmissions from the same event by the first one request from a gateway. Our extensional simulations verify the effectiveness of our proposed approach.

With the development of Location Based Services (LBSs), both academic researchers and industries have paid more attention to GPS-less mobile localization on mobile phones. The majority of the existing localization approaches have utilized signal-fingerprint as a metric for location determinations. However, one of the most challenging issues is the problem of uncertain fingerprints for building the fingerprint map, termed as the jigsaw puzzle problem. In this paper, for more accurate fingerprints of the mobile localization, we investigate the changes of Received Signal Strength Indication (RSSI) from the connected cell-towers over time along the mobile users' trajectories, termed as RSSI Time Series (RTS). Thus, we propose an RTS Assisted Localization System (RALS), which is a GPS-less outdoor mobile localization system. For localization, an RTS map is built on the back-end server, which consists of RTS harvested from the mobile phones, by the way of crowd sensing. The jigsaw puzzle problem slows down the map construction solely by the regular unintentional users with short-distance trajectories, and affects its efficiency. To speed up the map construction, we propose employing a few advanced intentional users with additional long-distance trajectories, at a higher cost than the regular user, this is called extra mile. Our extensional experiments verify the effectiveness of our localization system.

The development of the smartphones with various sensors, and powerful capabilities (computing, storage, and communication), motivates a popular computing and sensing paradigm, crowdsensing. In general, in crowdsensing, the smart-phones sense and collect the sensory data from a large number of smartphone users, for distinguishing the uncertain objects. However, some existing solutions for crowdsensing usually prefer to utilize only one or few features to distinguish the uncertain objects. In this paper, due to the limitation of less features, we propose to utilize multiple features to distinguish the uncertain objects for crowdsensing. For distinguishing uncertain objects with multiple features, we propose to utilize KL divergence based clustering. Moreover, we introduce two other mutated forms, the symmetry KL divergence and Jensen-Shannon KL divergence, to improve our algorithm. We evaluate our proposed schemes with real data of multiple features, which are collected by the smartphones with the sensors.

The data delivery in Vehicular Ad Hoc Networks (VANETs) depends on the mobility of the vehicles (e.g. with carry-and-forward). However, the mobility of the vehicles is not only affected by the nodes themselves, but also by some external means such as the traffic lights. The red light stops the vehicles at the intersection, which will increase the delivery delay of the messages carried by the vehicle with waiting time. On the contrary, this may also increase the opportunities of vehicles moving behind to catch up in forwarding messages. In this paper, we investigate the negative and positive influences of the traffic lights on data delivery in VANETs. We develop an analysis model for evaluating the data delivery among the vehicles that move along a path with multiple traffic lights. Based on the model, vehicles can estimate the reachability of destinations and the data delivery delay. Thus, we propose a transmission control scheme by the given deadline of reachable destinations, in order to improve the data delivery. Our intensive simulations verify the proposed model, and evaluate the influence of the traffic lights on data delivery.

Uncertain data are inherent in mobile crowd sensing applications, and the objects that they correspond to are usually vaguely specified. In order to improve performance, we often increase the number of features. However, the more features are used, the more redundancy and cost are involved correspondingly. Therefore, the number of features we selected for a specified application is a tradeoffs between the accuracy and the cost. In this paper, we model such tradeoffs between accuracy and cost as an optimization problem. Moreover, for investigating this problem, we propose to model the sensing with multiple features under a hypercube structure. In our scheme, each feature of uncertain objects is represented as a component of the vertex's coordinate in hypercube. At the same time, we prefer to define the edges between vertices with relative entropy rather than Euclidean distance. Because the former one could accurately measures the difference between two probability distributions of data. We evaluate our proposed schemes with real data of a crowd sensing recognition case, which are collected by smartphones with sensors.

Data delivery in Vehicular Ad Hoc Networks (VANETs) is based on the vehicles on the roads. However, the distribution of vehicles could be affected by some external means. For example, the traffic light or pedestrian signal could block the traffic flow moving onto a road. Thus, a gap between vehicles will appear at the entrance of the road, where the distance is larger than the communication range of the vehicles. We term it as a traffic hole, which not only affects the forwarding opportunities in VANETs, but also affects the performance of data delivery on the road, even under heavy traffic. In this paper, we model and analyze the traffic hole problem to characterize the pattern of traffic holes in VANETs. Then we discuss its influence on the data delivery in VANETs, and propose to utilize the backward traffic to mitigate the traffic hole problem. We conduct intensive simulations for discussing the traffic hole problem in VANETs. The simulation results imply that signal operations can affect the performance of data delivery in VANETs, and suggest that the backward traffic can mitigate the traffic hole problem.

Comfortable travel is an essential issue of Intelligent Transport Systems (ITS). However, the driver's behavior and the road condition affect the comfort of the passenger's riding experience while they are traveling. In this paper, we propose a system named Riding Experience Sensor (RESen) for sensing and evaluating the riding experience, based on crowd sourcing by smart phones. We utilize the acceleration sensor and gravity sensor for sensing with arbitrary orientations of smart phones. We partition the riding experience into horizontal and vertical for evaluation. Thus, based on the driver's historical trajectories, the system can provide feedbacks for improving driving by finding the anomalies along these trajectories. Based on the map, which has evaluated the comfort of each road, the system can provide a comfortable travel plan for query users.

Vehicular Ad Hoc Networks (VANETs) are getting increasing attention from academic researchers and automotive industries. Timely and cost-efficient multi-hop data delivery among vehicles is essential for VANETs, and various routing protocols are envisioned for infrastructure-less vehicle-to vehicle (V2V) communications. Due to the road-constrained data delivery and highly dynamic topology of vehicle nodes, it's better to construct routing based on the road-to-road pattern than the traditional node-to-node routing pattern in MANETs. However, the challenging issue for the road-to-road routing in VANETs is the opportunistic forwarding at intersections. Therefore, we propose a novel routing scheme, called Buffer and Switch (BAS). In BAS, each road buffers the data packets with multiple duplicates propagation in order to provide more opportunities for packet switching at intersections. Different from conventional protocols in VANETs, the propagation of duplicates in BAS is bidirectional along the routing path. Moreover, BAS's cost is much lower than other flooding-based protocols due to its spatio-temporally controlled duplicates propagation. We conduct the extensive simulations to evaluate the performance of BAS based on the road map of a real city collected from Google Earth. The simulation results show that BAS can outperform the existing protocols, especially when the network resources are limited.

In a wireless sensor network (WSN), the energy hole problem is a key factor which affects the lifetime of the networks. In a WSN with circular multi-hop deployment (modeled as concentric coronas), sensors in one corona have the same transmission range termed as the transmission range of this corona,