With the rapid development of Internet, the traditional computing environment is making a big migration to the cloud-computing environment. However, cloud computing introduces a set of new security problems. Aiming at the virtual machine (VM) escape attack, we study the traditional attack model and attack scenarios in the cloud-computing environment. In addition, we propose an access control model that can prevent virtual machine escape (PVME) by adapting the BLP (Bell-La Padula) model (an access control model developed by D. Bell and J. LaPadula). Finally, the PVME model has been implemented on full virtualization architecture. The experimental results show that the PVME module can effectively prevent virtual machine escape while only incurring 4% to 8% time overhead.
Live migration of virtual machines is an important approach for dynamic resource scheduling in cloud environment. The hybrid-copy algorithm is an excellent algorithm that combines the pre-copy algorithm with the post-copy algorithm to remedy the defects of the pre-copy algorithm and the post-copy algorithm. Currently, the hybrid-copy algorithm only copies all memory pages once in advance. In a write-intensive workload, copy memory pages once may be enough. However, more iterative copy rounds can significantly reduce the page faults in a read-intensive workload. In this paper, we propose a new parameter to decide the appropriate time to stop the iterative copy phase based on real-time situation. We use a Markov model to forecast the memory access pattern. Based on the predicted results and the analysis of the actual situation, the memory page transfer order would be adjusted to reduce the invalid transfers. The novel hybrid-copy algorithm is implemented on the Xen platform. The experimental results demonstrate that our mechanism has good performance both on read-intensive workloads and write-intensive workloads.
Endometrial polyps are common gynecological lesions. The standard treatment for this condition is hysteroscopic polypectomy. However, this procedure may be accompanied by misdetection of endometrial polyps. To improve the diagnostic accuracy and reduce the risk of misdetection, a deep learning model based on YOLOX is proposed to detect endometrial polyps in real time. Group normalization is employed to improve its performance with large hysteroscopic images. In addition, we propose a video adjacent-frame association algorithm to address the problem of unstable polyp detection. Our proposed model was trained on a dataset of 11,839 images from 323 cases provided by a hospital and was tested on two datasets of 431 cases from two hospitals. The results show that the lesion-based sensitivity of the model reached 100% and 92.0% for the two test sets, compared with 95.83% and 77.33%, respectively, for the original YOLOX model. This demonstrates that the improved model may be used effectively as a diagnostic tool during clinical hysteroscopic procedures to reduce the risk of missing endometrial polyps.
Provable Data Possession (PDP) protocol makes it possible for cloud users to check whether the cloud servers possess their original data without downloading all the data. However, most of the existing PDP schemes are based on either public key infrastructure (PKI) or identity-based cryptography, which will suffer from issues of expensive certificate management or key escrow. In this paper, we propose a new construction of certificateless provable group shared data possession (CL-PGSDP) protocol by making use of certificateless cryptography, which will eliminate the above issues. Meanwhile, by taking advantage of zero-knowledge protocol and randomization method, the proposed CL-PGSDP protocol leaks no information of the stored data and the group user’s identity to the verifiers during the verifying process, which is of the property of comprehensive privacy preservation. In addition, our protocol also supports efficient user revocation from the group. Security analysis and experimental evaluation indicate that our CL-PGSDP protocol provides strong security with desirable efficiency.
With the rapid development of video surveillance technology, especially the popularity of cloud-based video surveillance applications, video data begins to grow explosively. However, in the cloud-based video surveillance system, replicas occupy an amount of storage space. Also, the slow response to video playback constrains the performance of the system. In this paper, considering the characteristics of video data comprehensively, we propose a dynamic redundant replicas mechanism based on security levels that can dynamically adjust the number of replicas. Based on the location correlation between cameras, this paper also proposes a data cache strategy to improve the response speed of data reading. Experiments illustrate that: (1) our dynamic redundant replicas mechanism can save storage space while ensuring data security; (2) the cache mechanism can predict the playback behaviors of the users in advance and improve the response speed of data reading according to the location and time correlation of the front-end cameras; and (3) in terms of cloud-based video surveillance, our proposed approaches significantly outperform existing methods.