More Like this

1. Experimental Demonstration of Multipath and Continuous Repetition in Optical Packet/Circuit Networks

   https://doi.org/10.23919/OECC/PSC53152.2022.9850089  

   Hirota, Yusuke; Shiraiwa, Masaki; Xu, Sugang; Awaji, Yoshinari; Tornatore, Massimo; Mukherjee, Biswanath; Furukawa, Hideaki (July 2022, Proceedings of OECC)
2. Neural packet classification

   https://doi.org/10.1145/3341302.3342221  

   Liang, Eric; Zhu, Hang; Jin, Xin; Stoica, Ion (August 2019, ACM SIGCOMM)
3. Accelerating Packet Processing in Container Overlay Networks via Packet-level Parallelism

   https://doi.org/10.1109/IPDPS54959.2023.00018  

   Lei, Jiaxin; Munikar, Manish; Lu, Hui; Jia, Rao (May 2023, Proceedings IEEE International Parallel and Distributed Processing Symposium)

   Overlay networks serve as the de facto network virtualization technique for providing connectivity among distributed containers. Despite the flexibility in building customized private container networks, overlay networks incur significant performance loss compared to physical networks (i.e., the native). The culprit lies in the inclusion of multiple network processing stages in overlay networks, which prolongs the network processing path and overloads CPU cores. In this paper, we propose mFlow, a novel packet steering approach to parallelize the in-kernel data path of network flows. mFlow exploits packet-level parallelism in the kernel network stack by splitting the packets of the same flow into multiple micro-flows, which can be processed in parallel on multiple cores. mFlow devises new, generic mechanisms for flow splitting while preserving in-order packet delivery with little overhead. Our evaluation with both micro-benchmarks and real-world applications demonstrates the effectiveness of mFlow, with significantly improved performance – e.g., by 81% in TCP throughput and 139% in UDP compared to vanilla overlay networks. mFlow even achieved higher TCP throughput than the native (e.g., 29.8 vs. 26.6 Gbps). 

   more » « less
4. EPF: Evil Packet Filter

   Jin, Di; Atlidakis, Vaggelis; Kemerlis, Vasileios P. (July 2023, Proceedings of the USENIX Annual Technical Conference)

   The OS kernel is at the forefront of a system's security. Therefore, its own security is crucial for the correctness and integrity of user applications. With a plethora of bugs continuously discovered in OS kernel code, defenses and mitigations are essential for practical kernel security. One important defense strategy is to isolate user-controlled memory from kernel-accessible memory, in order to mitigate attacks like ret2usr and ret2dir. We present EPF (Evil Packet Filter): a new method for bypassing various (both deployed and proposed) kernel isolation techniques by abusing the BPF infrastructure of the Linux kernel: i.e., by leveraging BPF code, provided by unprivileged users/programs, as attack payloads. We demonstrate two different EPF instances, namely BPF-Reuse and BPF-ROP, which utilize malicious BPF payloads to mount privilege escalation attacks in both 32- and 64-bit x86 platforms. We also present the design, implementation, and evaluation of a set of defenses to enforce the isolation between BPF instructions and benign kernel data, and the integrity of BPF program execution, effectively providing protection against EPF-based attacks. Our implemented defenses show minimal overhead (<3%) in BPF-heavy tasks. 

   more » « less
5. Natural Graph Wavelet Packet Dictionaries

   https://doi.org/10.1007/s00041-021-09832-3  

   Cloninger, Alexander; Li, Haotian; Saito, Naoki (June 2021, Journal of Fourier Analysis and Applications)

   null (Ed.)

   Abstract We introduce a set of novel multiscale basis transforms for signals on graphs that utilize their “dual” domains by incorporating the “natural” distances between graph Laplacian eigenvectors, rather than simply using the eigenvalue ordering. These basis dictionaries can be seen as generalizations of the classical Shannon wavelet packet dictionary to arbitrary graphs, and do not rely on the frequency interpretation of Laplacian eigenvalues. We describe the algorithms (involving either vector rotations or orthogonalizations) to construct these basis dictionaries, use them to efficiently approximate graph signals through the best basis search, and demonstrate the strengths of these basis dictionaries for graph signals measured on sunflower graphs and street networks. 

   more » « less