Research publication catalog
Year-grouped research papers by Naser Ezzati-Jivan on kernel tracing, software performance, observability, systems, and machine learning.
Research profile · Catalog JSON · JSON Schema · Browse topics · Recurring keywords
89 publications
ACM/SPEC International Conference on Performance Engineering (ICPE)
B-Perf reconstructs execution, memory, and messaging behavior from Linux kernel-level traces and infers performance-antipattern indicators without requiring application source access or intrusive instrumentation.
conference paper full-text-reviewed
IEEE Transactions on Software Engineering
CARE combines distributed traces and profiling metrics with graph- and spectrum-based analysis to localize performance root causes in microservices.
journal article full-text-reviewed
Journal of Systems and Software
DTraComp is an open-source Eclipse Trace Compass framework that compares groups of distributed requests and attributes span time to user-space, kernel, thread-state, and system-call evidence.
journal article full-text-reviewed
Journal of Systems and Software
LMAT combines multi-task language models for kernel-event and event-duration prediction with online change detection, lightweight error-vector root-cause analysis, and adaptive tracing control.
journal article full-text-reviewed
Proceedings of the 34th ACM International Conference on the Foundations of Software Engineering (FSE)
The tutorial presents an end-to-end AgentOps workflow that connects task intent and model decisions to tool calls, memory access, inter-agent communication, external side effects, and runtime-governance decisions.
conference paper full-text-reviewed
SIGSOFT FSE Companion
The paper reframes performance debugging as collaborative reasoning over multiple evidence-grounded hypotheses rather than optimization for one supposedly best explanation.
conference paper full-text-reviewed
arXiv / CoRR; arXiv page reports acceptance to ICSE 2026
TAAF combines time-indexed knowledge graphs with large language models to turn large execution traces into queryable, actionable explanations.
preprint with publication signal full-text-reviewed
ACM International Conference on the Foundations of Software Engineering (FSE) Companion
TraceSynth generates novel structured kernel-trace windows with a Transformer diffusion model and repairs generated events against invariants mined from real LTTng traces.
conference paper full-text-reviewed
2025 International Conference on Artificial Intelligence, Computer, Data Sciences and Applications (ACDSA)
The paper implements a timestamp-aware multimodal video-retrieval pipeline that joins speech transcription, sampled-frame captioning, text embeddings, and approximate-nearest-neighbor search.
conference paper full-text-reviewed
IEEE International Conference on Program Comprehension (ICPC)
The paper introduces a taxonomy of log-event characteristics that explains why different log parsers fail across systems and parser families.
conference paper full-text-reviewed
Authorea preprint
The preprint combines communication-related kernel/system-call events with distributed traces to detect communication performance anti-patterns in microservices.
preprint full-text-reviewed
2025 IEEE International Conference on Collaborative Advances in Software and COmputiNg (CASCON)
The paper models LLM inference energy separately on CPU and GPU using hardware counters, device metrics, and task/model features, then compares classical and neural regressors across language tasks.
conference paper full-text-reviewed
IEEE/ACM International Conference on Software Engineering (ICSE)
The paper reconstructs missing system-call events in execution traces with diffusion and structured state-space generative models.
conference paper full-text-reviewed
IEEE International Conference on Software Maintenance and Evolution (ICSME)
HybridRCA combines critical-path-aware span analysis with targeted kernel metrics to reduce production trace volume while preserving root-cause localization evidence.
conference paper full-text-reviewed
AMCIS 2025, Data Science / SIG DSA (ERF)
The paper proposes the Triangle Multi-Dimensional Model for Bias Analysis, a hierarchical and interaction-based framework for tracing how bias originates, propagates, compounds, and feeds back across an LLM lifecycle.
conference paper full-text-reviewed
arXiv / CoRR
DEBA is an architecture-aware adaptive batch scheduler that uses training-stability signals to decide when and how to change batch size.
preprint full-text-reviewed
deep-learning-systems adaptive-scheduling performance-optimization
ACM/SPEC International Conference on Performance Engineering (ICPE)
The paper compares compression and optimization strategies for reducing the resource cost of Transformer and large-language-model workloads while retaining useful accuracy.
short conference paper full-text-reviewed
2025 IEEE International Conference on Collaborative Advances in Software and COmputiNg (CASCON)
SynthLogAI benchmarks statistical, sequence, transformer, and prompt-based generative models for producing synthetic Linux logs while measuring fidelity, downstream utility, and privacy.
conference paper full-text-reviewed
ACM/SPEC International Conference on Performance Engineering (ICPE)
The paper applies graph attention networks to predict future interactions in microservice call graphs, supporting proactive monitoring.
conference paper full-text-reviewed
Software Impacts
The paper proposes a two-tier IoT intrusion detector that first separates normal from anomalous flows and then classifies the attack type with a CNN-BiLSTM model.
journal article full-text-reviewed
ACM/SPEC International Conference on Performance Engineering (ICPE)
ALS uses source-code features and reinforcement learning to recommend which Python functions to log and which log levels to use for performance-bug diagnosis.
conference paper full-text-reviewed
ACM/SPEC ICPE Companion
The paper identifies response-time variability in Alibaba microservice traces through critical-path extraction and variability analysis of service interactions.
conference paper full-text-reviewed
2024 IEEE International Conference on Big Data (BigData)
The paper evaluates whether software-energy predictors transfer across applications when they use hardware-performance and operating-system event representations rather than application-specific measurements alone.
conference paper full-text-reviewed
energy-efficiency performance-modeling predictive-monitoring
The 37th Canadian Conference on Artificial Intelligence
The paper uses gradient boosting and feature importance to reduce the amount of trace data needed for microservice performance modeling.
conference paper full-text-reviewed
ACM/SPEC ICPE Companion
The work-in-progress paper combines service-call graph context, social-network analysis, and spectrum-based fault localization to rank distributed-trace root causes.
work-in-progress conference paper full-text-reviewed
2024 ACM/IEEE International Conference on Software Engineering: Companion Proceedings (ICSE Companion)
The paper derives a 30-item taxonomy of log event characteristics that induce parsing errors and quantifies the characteristics with the largest impact across eight parsers.
poster paper full-text-reviewed
ACM/SPEC International Conference on Performance Engineering (ICPE) Companion
The paper combines an unsupervised GraphSAGE model with critical-path-specific latency profiles to detect anomalous requests and rank likely microservice culprits without labelled training data.
conference paper full-text-reviewed
Journal of Systems and Software
The paper releases a reusable Linux kernel-event and system-call artifact with controlled application workloads, injected resource noise, and analysis scenarios for software performance engineering.
journal article full-text-reviewed
2024 IEEE International Conference on Collaborative Advances in Software and COmputiNg (CASCON)
MemAdapt forecasts memory behavior under irregular sampling and uses the forecast to choose an adaptive monitoring rate that balances estimation quality with collection overhead.
conference paper full-text-reviewed
resource-analysis predictive-monitoring performance-analysis
ACM/SPEC International Conference on Performance Engineering (ICPE) Companion
The paper applies graph community detection and service-graph clustering to expose recurring microservice communication structures in an Alibaba production-cluster snapshot.
conference paper full-text-reviewed
ACL 2024 Main Conference
The paper introduces WinoVis, a multimodal benchmark and analysis framework for testing pronoun disambiguation in text-to-image models.
conference paper full-text-reviewed
2024 IEEE International Conference on Collaborative Advances in Software and COmputiNg (CASCON)
The paper detects and localizes performance noise by aligning passive kernel-event evidence with the life cycle of CPU, disk, and network requests rather than treating resource utilization as an undifferentiated aggregate.
conference paper full-text-reviewed
2024 IEEE/ACM International Conference on Software Engineering: New Ideas and Emerging Results (ICSE-NIER)
The paper proposes an adaptive tracing loop that combines language-model prediction of kernel-event sequences and event durations with change detection and root-cause analysis, so detailed tracing is activated only around significant behavior shifts.
conference paper full-text-reviewed
IEEE International Conference on Big Data
AltOOM combines early memory-pressure forecasting with selective process-level profiling to identify the process most responsible for an impending out-of-memory event.
conference paper full-text-reviewed
ACM/IFIP/USENIX Middleware 2023 Industry Track
The paper applies a CNN-BiLSTM intrusion classifier to RPL/IoT traffic, combining convolutional feature extraction with bidirectional sequence modeling after imbalance-aware flow preprocessing.
industry-track paper full-text-reviewed
2023 IEEE International Conference on Trust, Security and Privacy in Computing and Communications (TrustCom)
EMD-SCS combines sequence prediction of system calls with sonification so that partial execution prefixes can support early malware detection and an interpretable auditory alert.
conference paper full-text-reviewed
ACM/SPEC ICPE Companion
The paper uses time-series analysis to identify and classify warmup, steady-state, and anomalous states in Java JMH microbenchmarks.
conference paper full-text-reviewed
performance-analysis performance-engineering anomaly-detection
2023 IEEE International Conference on Software Engineering and Research (SERA)
The paper presents a multi-level adaptive tracing workflow that uses lightweight stress detection and performance evidence to change the application/kernel instrumentation scope during an investigation.
conference paper full-text-reviewed
IEEE Working Conference on Source Code Analysis and Manipulation (SCAM)
PASD uses Linux kernel-event traces, Perf call stacks, and statistical debugging to rank functions associated with performance problems without application-source instrumentation.
conference paper full-text-reviewed
Companion of the 2023 ACM/SPEC International Conference on Performance Engineering (ICPE '23 Companion)
The study examines whether static source-code features are associated with the steady-state behavior of Java microbenchmarks during JVM warmup.
conference paper full-text-reviewed
performance-analysis performance-engineering machine-learning
IEEE International Conference on Program Comprehension (ICPC)
The paper classifies recurring log-parsing errors and connects them to token and event characteristics that can guide parser design and logging practice.
conference paper full-text-reviewed
ACM/SPEC ICPE Companion
The paper uses sampled call graphs and machine learning to distinguish warmup from steady-state executions in Java microbenchmarks.
conference paper full-text-reviewed
performance-analysis performance-engineering machine-learning
ACM/IFIP/USENIX Middleware 2022 Demos/Posters
The poster proposes a hybrid IoT intrusion-detection pipeline with binary anomaly detection followed by multiclass attack classification.
poster abstract full-text-reviewed
Concurrency and Computation: Practice and Experience
The paper combines model-based quantitative constraints with user-space and kernel execution traces to automatically identify likely causes of real-time and multicore constraint violations.
journal article full-text-reviewed
IEEE Working Conference on Source Code Analysis and Manipulation (SCAM)
N-Lane Bridge extends One-Lane Bridge analysis to active-resource bottlenecks and uses LTTng system-level traces to distinguish application-caused congestion from external interference.
conference paper full-text-reviewed
30th International Conference on Program Comprehension (ICPC)
The framework combines critical-path extraction, coarse anomaly checks, OPTICS execution clustering, and MUSCLE multiple sequence alignment to localize performance differences in system-level traces.
conference paper full-text-reviewed
Performance Evaluation
The paper unifies Chromium user-space events with Linux kernel scheduling and resource events to analyze execution paths and localize sporadic latency in a complex multithreaded application.
journal article full-text-reviewed
2022 IEEE International Conference on Cloud Engineering (IC2E)
This poster proposes operating-system-aware antipattern detection that uses system-level traces and scalability trends to expose resource blocking that aggregate CPU utilization can hide.
poster paper full-text-reviewed
IEEE International Conference on Software Quality, Reliability and Security (QRS)
The paper combines system-level dependency graphs with automated outlier detection to localize likely causes of latency anomalies.
conference paper full-text-reviewed
International Journal of Parallel Programming
The paper automatically builds an approximate workflow model and adaptive quantitative constraints from user-space and kernel execution traces, reducing manual model construction for real-time and multicore diagnosis.
journal article full-text-reviewed
Electronics
The paper combines Jaeger distributed traces with LTTng kernel and user-space events and extends critical-path analysis from threads to distributed requests so request latency can be related to low-level scheduling and resource behavior.
journal article full-text-reviewed
IEEE International Conference on Cloud Computing Technology and Science (CloudCom)
The paper characterizes container workloads from host-level LTTng traces by selecting important threads and clustering execution-state behavior without instrumenting the containers internally.
conference paper full-text-reviewed
Wireless Communications and Mobile Computing
The paper correlates LTTng traces from user space through kernel, storage, network, and multiple hosts in a disk-backed state model, enabling top-down diagnosis of distributed request latency.
journal article full-text-reviewed
Sensors
The paper presents a modular client-server architecture for large-trace storage, analysis, and visualization inside an integrated development environment.
journal article full-text-reviewed
2021 IEEE International Conference on Big Data (BigData)
The paper presents a low-intrusion heap monitor that records allocation call stacks and uses persistent heap-growth filtering plus trace correlation to identify likely memory-leak roots.
conference paper full-text-reviewed
2021 IEEE International Symposium on Networks, Computers and Communications (ISNCC)
The paper provides a graphical state-machine modeling tool that generates trace analyses and integrates them into Trace Compass, reducing the need to hand-code state providers or XML analyses.
conference paper full-text-reviewed
2021 IEEE International Systems Conference (SysCon)
The paper combines LTTng system-call monitoring with recurrent and Transformer-style models for malware and anomalous system-behavior detection.
conference paper full-text-reviewed
2021 IEEE International Systems Conference (SysCon)
The work bridges kernel-level hardware-counter sampling and user-space interpreter state so execution-path profiling can attribute performance evidence to interpreted code.
conference paper full-text-reviewed
Journal of Hardware and Systems Security
The paper proposes a cloud-side anomaly detector that combines traffic, virtual-machine resource, and kernel-level indicators to identify several DDoS classes through change-point evidence.
journal article full-text-reviewed
IEEE Internet of Things, GreenCom, CPSCom and SmartData
The paper correlates LTTng traces across OpenStack services, QEMU/KVM, network components, and host kernels to diagnose cloud-platform performance problems.
conference paper full-text-reviewed
IEEE 20th International Working Conference on Source Code Analysis and Manipulation (SCAM)
DepGraph uses system-level tracing to expose the waiting dependencies that divide a task's execution time across threads and hardware resources.
conference paper full-text-reviewed
Softwaretechnik-Trends
The paper adapts community detection and PageRank from social-network analysis to reduce and prioritize system execution traces.
article full-text-reviewed
IEEE International Symposium on Software Reliability Engineering Workshops (ISSREW)
The paper extends critical-path analysis across kernel and user-space lock events so contention from futexes, spinlocks, semaphores, and other synchronization mechanisms can be diagnosed without recompiling the application.
conference paper full-text-reviewed
Journal of Cloud Computing
The paper combines host-level user/kernel tracing with machine learning to detect intrusion behavior on constrained IoT and home-automation systems.
journal article full-text-reviewed
IEEE International Symposium on Software Reliability Engineering Workshops (ISSREW)
The paper presents a tracing and analysis pipeline that detects slow web requests and groups them by internal execution behavior to expose likely causes.
workshop paper full-text-reviewed
Journal of Systems Architecture
The paper redesigns GDB dynamic tracing around LTTng-UST and scalable trace views so multi-threaded CPU/GPU debugging remains usable on heterogeneous many-core systems.
journal article full-text-reviewed
International Journal of Parallel Programming
The paper partitions Common Trace Format streams into balanced workloads, resolves cross-chunk state dependencies, and parallelizes stateful trace analyses with low synchronization cost.
journal article full-text-reviewed
Journal of Hardware and Systems Security
The paper combines execution-trace and virtual-machine metrics with machine learning to detect EDoS behavior and restrict resource expansion to apparently normal VMs.
journal article full-text-reviewed
IEEE International Conference on Consumer Electronics (ICCE)
The paper correlates OpenStack Nova, QEMU/KVM, and host/guest kernel evidence to diagnose live-migration behavior across cloud-system layers.
conference paper full-text-reviewed
IEEE Systems Conference
The paper correlates PHP user-space events with LTTng kernel events in a unified multilevel model to explain high-latency web requests.
conference paper full-text-reviewed
Software Quality Journal
The framework lets analysts define stateful trace analyses declaratively, using a generic state model and XML specifications that can drive storage, filtering, and visualization across trace formats.
journal article full-text-reviewed
IEEE High Performance Extreme Computing Conference (HPEC)
The paper introduces a kernel-resident dynamic sampler that tracks virtual and physical memory usage while reducing the event volume generated by high-frequency memory activity.
conference paper full-text-reviewed
Concurrency and Computation: Practice and Experience
The paper designs and evaluates hardware-assisted alternatives to LTTng-UST for software-event and execution-path tracing, including STM/ETB, ETM/ETB, and a modified Perf/BTS path.
journal article full-text-reviewed
Concurrency and Computation: Practice and Experience
The survey provides a taxonomy and requirements-oriented comparison of techniques for collecting, abstracting, analyzing, visualizing, and navigating large execution traces.
survey article full-text-reviewed
IEEE BigData Congress
eSHT redesigns the State History Tree to store and query highly parallel trace intervals more compactly and with shallower trees.
conference paper full-text-reviewed
IEEE Future Internet of Things and Cloud Workshops (FiCloudW)
The paper fuses synchronized host and virtual-machine kernel traces into a Fused Virtual Machine view that exposes how VM threads, virtual CPUs, hypervisor transitions, and physical CPUs interact.
conference paper full-text-reviewed
IEEE BigData Congress
The paper provides a declarative XML/state-machine approach for flexible filtering and synthetic-event generation over large kernel traces.
conference paper full-text-reviewed
Concurrency and Computation: Practice and Experience
The paper proposes a multilevel data-cube model for computing statistics over live execution traces across several time scales.
journal article full-text-reviewed
2014 IEEE 27th Canadian Conference on Electrical and Computer Engineering (CCECE)
The paper presents multiscale trace navigation that links raw kernel events to system-call, synthetic, and fault/alert abstractions in a zoomable timeline.
conference paper full-text-reviewed
ACM SIGOPS Operating Systems Review
The framework computes system-parameter statistics for arbitrary intervals and resource hierarchies from very large traces using a disk-resident history structure and controlled granularity.
journal article full-text-reviewed
Runtime Verification 2013 (RV 2013)
The paper introduces a disk-backed state-history model that incrementally converts trace events into queryable state intervals, enabling scalable timestamp/state queries and visualization without replaying the entire trace.
conference paper full-text-reviewed
2013 IEEE 26th Canadian Conference on Electrical and Computer Engineering (CCECE)
The paper develops a multilevel trace-label placement algorithm that preserves readable, semantically meaningful labels while users zoom and pan across dense execution-trace views.
conference paper full-text-reviewed
IEEE International Conference on Social Computing
The paper presents a disk-oriented State History Tree that stores very large, time-ordered trace intervals without repeated node rebalancing.
conference paper full-text-reviewed
Advances in Software Engineering
The paper introduces a stateful trace-abstraction method that converts low-level kernel events into semantic and synthetic events using modeled resource state and finite-state patterns.
journal article full-text-reviewed
16th IASTED International Conference on Software Engineering and Applications (SEA 2012)
The paper introduces a stateful trace-abstraction architecture that shares modeled system state and intermediate computation across concurrent patterns to reduce the cost of analyzing large Linux kernel traces.
conference paper full-text-reviewed
International Journal of Computer Science and Network Security
This survey organizes intrusion-response systems around automation, response cost, decision evaluation, adjustment, execution, attack paths, and risk assessment.
journal article full-text-reviewed
Sixth International Conference on Internet Monitoring and Protection (ICIMP 2011)
The paper moves part of an application proxy into the kernel to reduce context switches and unnecessary data copies during connection filtering.
conference paper full-text-reviewed
performance-engineering performance-optimization resource-analysis
Wireless Sensor Network
The paper proposes L-SYNCng, a time-synchronization protocol for heterogeneous wireless sensor networks that combines topology-aware clustering with convex-hull clock estimation.
journal article full-text-reviewed
International Symposium on Information Management in a Changing World (IMCW 2010), Communications in Computer and Information Science 96
The paper combines automated Persian-news categorization with a web system for retrieving similar news items.
conference chapter full-text-reviewed