Welcome to FOCI
FOCI is a non-profit, open-access academic journal strictly dedicated to the rapid dissemination of research in Computer Science and related technological fields. We prioritize technical excellence, algorithmic efficiency, and mathematical rigor.
Recent Publications
The Latency Tax: Mitigating Stop-The-World Garbage Collection Pauses in High-Throughput Distributed Web Servers via Off-Heap Arena Architectures
A. Lin, J. Peterson, K. Zhang | Vol 1, Issue 1
This paper analyzes the performance degradation caused by GC pauses in modern backend infrastructure, where the cumulative probability of hitting a stop-the-world event increases exponentially across distributed microservice call chains. We propose an off-heap memory arena architecture that segregates long-lived persistent state from ephemeral request data, making millions of objects invisible to the garbage collector. Under sustained load of 50,000 RPS on a distributed key-value cluster, our implementation reduces 99th percentile latency by 42.1% (from 145ms to 84ms), 99.9th percentile latency by 71.9% (from 410ms to 115ms), and improves maximum throughput by 23.1%, with arena reset operations executing in O(1) time. We conclude that the latency reduction justifies the architectural complexity for high-throughput distributed web servers.
View Full Paper →Zero-Knowledge Proofs for State Verification in Decentralized Networks
M. Kovac, R. Al-Fayed | Vol 1, Issue 1
We present a novel implementation of zk-SNARKs tailored for verifying state transitions in distributed ledgers without exposing underlying transactional data, achieving a 15% reduction in computational overhead compared to baseline Groth16 implementations while maintaining 288-byte proofs and 18ms verification latency on Ethereum Layer 1.
View Full Paper →Lock-Free Data Structures in Real-Time Operating Systems
D. Rossi, E. Chen | Vol 1, Issue 1
An analysis of lock-free queue implementations using Compare-And-Swap (CAS) operations within the context of strict real-time OS schedulers, preventing priority inversion.
View Full Paper →