02 / Research topics
Potential PhD Research Topics in Cloud Computing
Explore suggested topics and the methods, tools and expertise that could support each investigation.
1
Carbon Aware Scheduling for Sustainable Multi Cloud Computing Infrastructures
Although cloud providers publish sustainability goals, workload schedulers rarely coordinate carbon intensity, latency, cost, and reliability across regions and providers. This research can develop carbon-aware prediction and optimization methods, quantify environmental tradeoffs, and produce an open scheduling framework validated with real traces and cloud testbeds. Novelty should address intermittent renewable energy, embodied emissions, and service-level constraints rather than optimizing electricity alone.
Applications and use cases
2
Confidential Federated Learning Across Heterogeneous Multi Cloud and Edge Environments
Develop privacy-preserving federated learning that remains accurate and efficient when participants, networks, hardware, and cloud providers differ substantially. The gap lies in jointly handling poisoning attacks, non-independent data, communication costs, and regulatory isolation. Objectives include adaptive aggregation, confidential computation, and robust incentives, with evaluation against centralized and conventional federated baselines using realistic cross-silo and edge datasets.
Applications and use cases
3
Autonomous AIOps for Predictive Failure Management in Cloud Native Platforms
Investigate why current AIOps systems generate excessive alerts and recommend actions without dependable causal evidence. The proposed contribution can combine topology-aware observability, causal inference, and safe reinforcement learning to predict failures and automate remediation. Objectives should include explainable diagnosis, uncertainty estimation, rollback protection, and reproducible comparisons using production-like microservice faults, emphasizing operator trust and avoided outages.
Applications and use cases
4
Verifiable Security Policies for Zero Trust Cloud Native Application Supply Chains
Address the gap between declarative zero-trust policies and their actual enforcement across identities, workloads, artifacts, and runtime networks. This PhD could create a formally analyzable policy language, automated verification methods, and continuous provenance controls for software supply chains. Evaluation should measure policy completeness, attack resistance, false positives, operational overhead, and usability against realistic dependency-confusion and compromised-build scenarios.
Applications and use cases
5
Digital Twins for Resilient and Self Optimizing Cloud Data Center Operations
Design digital twins that represent compute, network, cooling, energy, and workload dependencies sufficiently well for safe operational decisions. Existing models often optimize isolated resources and become inaccurate under changing workloads. The research should create continuously calibrated twins, test predictive and what-if control strategies, and demonstrate improvements in resilience, utilization, energy efficiency, and recovery time without risking production infrastructure.
Applications and use cases
6
Interpretable Serverless Computing Optimization Under Cold Start and Cost Uncertainty
Examine how serverless applications can achieve predictable latency and cost when cold starts, bursty demand, provider limits, and opaque billing interact. A novel approach may combine interpretable workload forecasting, adaptive placement, and multi-provider orchestration. Objectives include explaining optimization decisions, reducing tail latency and waste, and validating generalization across languages, runtimes, event patterns, and commercial serverless platforms.
Applications and use cases
7
Trustworthy Generative AI Services for Secure Cloud Software Engineering
Create methods for deploying generative AI coding and operations assistants without amplifying insecure code, data leakage, hallucinated actions, or license violations. The research gap concerns end-to-end assurance across prompts, models, tools, repositories, and runtime environments. Objectives can include retrieval controls, policy-aware agents, provenance tracking, adversarial testing, and human-centered evaluation against security, productivity, and reliability baselines.
Applications and use cases
8
Edge Cloud Continuum Orchestration for Real Time Intelligent Internet Applications
Evaluate how applications should place models, data, and services across devices, edge nodes, and centralized clouds under mobility, limited bandwidth, and strict latency requirements. Current orchestration largely treats edge resources as small clouds and ignores context and model adaptation. The proposal should deliver context-aware placement and migration algorithms, quantify quality-of-service tradeoffs, and validate them with realistic mobility and failure scenarios.
Applications and use cases
9
Post Quantum Cryptographic Migration Strategies for Long Lived Cloud Services
Unify cryptographic inventory, dependency analysis, performance measurement, and migration planning for cloud services exposed to harvest-now-decrypt-later threats. Existing guidance is high-level and rarely quantifies application-specific disruption across certificates, APIs, storage, and distributed identities. Objectives include automated discovery, hybrid protocol evaluation, risk prioritization, and migration playbooks tested across heterogeneous cloud architectures and realistic operational constraints.
Applications and use cases
10
Fair and Accountable Resource Allocation in Shared Cloud Computing Markets
Advance cloud resource allocation beyond utilization and revenue by addressing fairness between tenants, providers, geographic regions, and application priorities. The research can formulate transparent allocation mechanisms incorporating performance, carbon impact, accessibility, and contractual rights. Contributions should establish measurable fairness definitions, detect discriminatory outcomes, design incentive-compatible algorithms, and test them with traces, simulations, and stakeholder-informed scenarios.
Applications and use cases