FPGAs are widely used for efficient CNN inference acceleration but designing high-performance accelerators demands significant hardware expertise. Existing solutions face limitations: hardware designs are often model/chip-specific with suboptimal resource efficiency, and compiler support is typically framework-restricted. To overcome these, we propose a generalized and flexible high-performance FPGA accelerator architecture and a flexible end-to-end compilation toolflow based on ONNX IR. The architecture features an optimized uint8 systolic array for high compute density and a dedicated X-bus module handling diverse convolution parameters. On-chip buffers and allocation algorithms enhance memory efficiency. Configurable design variables enable architectural adaptation and fine-tuning. Deploying four accelerator variants on a VCU118 board and compiling 17 CNN models demonstrated a peak convolutional throughput of 5,792.19 GOPS (99.82% of theoretical peak, 5,825.42 GOPS) and overall throughput up to 3,311.48 GOPS. Compared to prior work, our solution offers superior usability, greater flexibility, and higher performance under comparable DSP usage. Furthermore, across most tested models, it provides significantly lower latency and higher energy efficiency versus CPUs and GPUs.

Unmanned aerial vehicles (UAVs) have opened new opportunities for crowd-sensing enabling the execution of sensing tasks in remote or rural regions by leveraging data sensing and computation offloading capabilities. However, UAVs often lack the incentive to actively engage in crowd-sensing tasks. Auction has been proven as an effective strategy to boost participant motivation and engagement. By incentivizing UAV owners, high-quality crowd-sensing tasks can be completed through increased task completion rates and improved data quality achieved by UAVs. In this paper, we formally model this QoE-aware UAV-enabled crowd-sensing problem as an optimization problem with the aim of maximizing the overall Quality of Experience (QoE) under resource and budget constraints. To solve this problem, we propose a QoE-aware online auction algorithm named CERA, based on the primal-dual technique and theoretically prove its competitive ratio. We conducted both small-scale and large-scale experiments to evaluate the performance of our approach, and the experimental results demonstrate that our CERA outperforms other algorithms significantly.

With the rapid increase in the use of mobile devices in people’s daily lives, mobile data traffic is exploding in recent years. In the edge computing environment where edge servers are deployed in close proximity to mobile users, caching popular data on edge servers can ensure mobile users’ low-latency access to those data and reduce the data traffic between mobile users and the centralized cloud. Existing studies consider the data caching problem with a focus on the reduction of network delay and the improvement of mobile devices’ energy efficiency. In this article, we tackle this data caching problem in the edge computing environment from a service provider’s perspective with the aim to maximize its data caching revenue. This problem is challenging because there is a trade-off between the benefit produced and the cost incurred by caching data on edge servers. In the meantime, the constraint for data access latency must also be fulfilled. In this article, we formulate the data caching problem in the edge computing environment as an integer programming (IP) problem and prove its NP-completeness. To solve this problem effectively and efficiently in large-scale scenarios, we propose an approximation approach to find near-optimal solutions. Extensive experiments are conducted on a widely-used real-world dataset to evaluate our approaches.

Edge data caching has attracted tremendous attention in recent years. Service providers can consider caching data on nearby locations to provide service for their app users with relatively low latency. The key to enhance the user experience is appropriately choose to cache data on the suitable edge servers to achieve the service providers' objective, e.g., minimizing data retrieval latency and minimizing data caching cost, etc. However, Quality of Experience (QoE), which impacts service providers' caching benefit significantly, has not been adequately considered in existing studies of edge data caching. This is not a trivial issue because QoE and Quality-of-Service (QoS) are not correlated linearly. It significantly complicates the formulation of cost-effective edge data caching strategies under the caching budget, limiting the number of cache spaces to hire on edge servers. We consider this problem of QoE-aware edge data caching in this paper, intending to optimize users' overall QoE under the caching budget. We first build the optimization model and prove the NP-completeness about this problem. We propose a heuristic approach and prove its approximation ratio theoretically to solve the problem of large-scale scenarios efficiently. We have done extensive experiments to demonstrate that the MPSG algorithm we propose outperforms state-of-the-art approaches by at least 68.77%.

Mobile data traffic is exploding in recent years with the exponential growth of mobile users. In the edge computing environment where edge servers are deployed around mobile users, caching data on edge servers can ensure mobile users' fast access to those data and reduce the data traffic between mobile users and the centralized cloud. Existing studies consider the data cache and replacement problem with the consideration of the reduction of network delay and the improvement of mobile devices' energy efficiency. In this paper, we attack proactive data caching and replacement problem in the edge computing environment from the service providers' perspective, who would like to maximize their venues of caching their data. This problem is complicated because data caching produces benefits at a cost and there usually is a trade-off in-between. In this paper, we formulate the data caching and replacement problem as an integer programming problem, and maximizes the revenue of the service provider while satisfying a constraint for data access latency. We also propose an online algorithm to solve problems in large-scale scenarios. Extensive experiments are conducted on a real-world dataset that contains the locations of edge serPvers and mobile users.

With the rapid increase in the use of mobile devices in people's daily lives, mobile data traffic is exploding in recent years. In the edge computing environment where edge servers are deployed around mobile users, caching popular data on edge servers can ensure mobile users' fast access to those data and reduce the data traffic between mobile users and the centralized cloud. Existing studies consider the data cache problem with a focus on the reduction of network delay and the improvement of mobile devices' energy efficiency. In this paper, we attack the data caching problem in the edge computing environment from the service providers' perspective, who would like to maximize their venues of caching their data. This problem is complicated because data caching produces benefits at a cost and there usually is a trade-off in-between. In this paper, we formulate the data caching problem as an integer programming problem, and maximizes the revenue of the service provider while satisfying a constraint for data access latency. Extensive experiments are conducted on a real-world dataset that contains the locations of edge servers and mobile users, and the results reveal that our approach significantly outperform the baseline approaches.

Software as a service (SaaS) is an emerging software framework in which business data and logic typically integrate with other applications. It requires a unified subscriber to describe SaaS to make for easy integration, however, SaaS provides services to different tenants by running only one instance. In order to satisfy personalized needs from different tenants, the business logic becomes correspondingly complex. As this logic is cumbersome to reveal to every individual tenant, we propose the use of Web Services Conversation Language (WSCL) to express the views of tenant and provider separately. To overcome deficiencies in WCSL for expressing heterogeneous data, process rules, and business rules, we extend the syntax of WSCL. We also put forward a new modeling method for constructing SaaS Service, describing the modeling process and the algorithm for obtaining the tenant model from the business model. In conclusion, we describe the modeling tools and validation methods.

With the increasing numbers of Web services and service users on World Wide Web, predicting QoS(Quality of Service) for users will greatly aid service selection and discovery. Due to the different backgrounds and experiences of users, they have different QoS experiences when interacting with the same service. Even two users who have similar experiences on some services can have diverging views when considering other services. This paper proposes an approach to predict QoS. It is based on not only other users' QoS experiences, but also the environment factor and user input factor. First bring forwards usage information feature model and calculate the similarity of two users based on the feature model. Then consider not only the historic information, but also environment and users' inputs, such as bandwidth and data size. Before calculating the user similarity, select a set of Web services that have the highest degree of similarity with the target service, not all of the services. The missing value can be calculated through the data of similar services. The results of the experiment prove that our approach is feasible and effective.