Publiée 22 juin 2026
PhD Position F/M Concurrent Autonomic Control for the Computing Continuum
Inria
Paris, Île-de-France 75000, France
CDI
Rémunération: 2 300 $ Hebdomadaire
A propos du centre ou de la direction fonctionnelle
This position is situated in Paris, but offered by the Inria research centre in Lyon.
The Inria research centre in Lyon is the 9th Inria research centre, formally created in January 2022. It brings together approximately 410 people in 20 research teams and research support services.
Its staff are distributed in Villeurbanne, Lyon Gerland, and Saint-Etienne.
The Lyon centre is active in the fields of software, distributed and high-performance computing, embedded systems, quantum computing and privacy in the digital world, but also in digital health and computational biology.
Contexte et atouts du poste
The PhD will take place at the Research Center of the Léonard de Vinci Pôle Universitaire (DVRC), in Paris La Défense. It will be jointly supervised by Christian PEREZ (AVALON team, Inria, at the LIP laboratory) and by Farah AIT-SALAHT (ESILV/DVRC). The PhD candidate will have access to the Grid'5000 and SLICES platforms and to the tools developed by Taranis project.
This thesis involves several academic partners within the TARANIS project of PEPR Cloud; the PhD will therefore be carried out in a national collaboration context. Annual meetings in other cities in France as well as regular visits to the AVALON team (ENS Lyon) are to be expected.
This PhD will be carried out within the framework of the France 2030 programme and the Programmes et Équipements Prioritaires de Recherche (PEPR) Cloud , specifically the TARANIS project and its Work Package 3 on Service and Resource Orchestration. The Computing Continuum (Cloud-Edge-IoT) has become the reference infrastructure for deploying large-scale distributed applications, but it is deeply heterogeneous, distributed, and dynamic by nature. These properties raise major challenges for resource management and service orchestration. To address them, autonomic systems, capable of self-configuration, self-optimization, and self-healing, rely on the MAPE-K reference architecture (Monitor, Analyze, Plan, Execute over a shared knowledge base), which can be instantiated through a variety of patterns (single loop, hierarchical, coordinated, peer-to-peer).
Most existing MAPE-K instantiations adopt a strictly sequential ordering of the four phases, which introduces a significant cumulative latency and degrades responsiveness to rapid changes (load spikes, failures, mobility). In the context of service placement, where deployment and migration decisions must meet Quality-of-Service constraints, this latency becomes a critical bottleneck. Existing solutions (decentralized multi-loop architectures, optimization of individual phases) distribute or accelerate some phases of the MAPE-K loop but do not break the fundamental sequential dependency between the planning and execution phases within a single loop.
Mission confiée
This thesis aims to improve the MAPE-K loop by investigating the overlap between the Plan and Execute phases, possibly running several P-E cycles in parallel, under explicit consistency and conflict-resolution guarantees. More precisely, the objectives of the thesis is to design, formalize, and evaluate a concurrent coordination pattern for MAPE-K, referred to as MACPE-K, for service orchestration in the Cloud-Edge-IoT continuum. The work will formally establish the conditions under which the Plan and Execute phases can overlap while guaranteeing consistency and resolving action-level conflicts, and then model the end-to-end responsiveness/cost trade-off, including the Monitoring and Analysis phases. The dynamic tuning of the degree of concurrency and of the triggering policies will be investigated in order to preserve responsiveness under varying workloads and environmental conditions. Service placement will serve as the primary case study, but a second self-management task (e.g., elastic autoscaling or fault recovery) will also be instantiated to substantiate the genericity of the pattern, covering both stateless and stateful services
Principales activités
The work of the PhD will be organized in several phases:
Compétences
We are looking for a candidate holding (or about to obtain) a Master's degree or engineering diploma in computer science or a related field, with:
Avantages
Rémunération
2300 euros gross salary /month
This position is situated in Paris, but offered by the Inria research centre in Lyon.
The Inria research centre in Lyon is the 9th Inria research centre, formally created in January 2022. It brings together approximately 410 people in 20 research teams and research support services.
Its staff are distributed in Villeurbanne, Lyon Gerland, and Saint-Etienne.
The Lyon centre is active in the fields of software, distributed and high-performance computing, embedded systems, quantum computing and privacy in the digital world, but also in digital health and computational biology.
Contexte et atouts du poste
The PhD will take place at the Research Center of the Léonard de Vinci Pôle Universitaire (DVRC), in Paris La Défense. It will be jointly supervised by Christian PEREZ (AVALON team, Inria, at the LIP laboratory) and by Farah AIT-SALAHT (ESILV/DVRC). The PhD candidate will have access to the Grid'5000 and SLICES platforms and to the tools developed by Taranis project.
This thesis involves several academic partners within the TARANIS project of PEPR Cloud; the PhD will therefore be carried out in a national collaboration context. Annual meetings in other cities in France as well as regular visits to the AVALON team (ENS Lyon) are to be expected.
This PhD will be carried out within the framework of the France 2030 programme and the Programmes et Équipements Prioritaires de Recherche (PEPR) Cloud , specifically the TARANIS project and its Work Package 3 on Service and Resource Orchestration. The Computing Continuum (Cloud-Edge-IoT) has become the reference infrastructure for deploying large-scale distributed applications, but it is deeply heterogeneous, distributed, and dynamic by nature. These properties raise major challenges for resource management and service orchestration. To address them, autonomic systems, capable of self-configuration, self-optimization, and self-healing, rely on the MAPE-K reference architecture (Monitor, Analyze, Plan, Execute over a shared knowledge base), which can be instantiated through a variety of patterns (single loop, hierarchical, coordinated, peer-to-peer).
Most existing MAPE-K instantiations adopt a strictly sequential ordering of the four phases, which introduces a significant cumulative latency and degrades responsiveness to rapid changes (load spikes, failures, mobility). In the context of service placement, where deployment and migration decisions must meet Quality-of-Service constraints, this latency becomes a critical bottleneck. Existing solutions (decentralized multi-loop architectures, optimization of individual phases) distribute or accelerate some phases of the MAPE-K loop but do not break the fundamental sequential dependency between the planning and execution phases within a single loop.
Mission confiée
This thesis aims to improve the MAPE-K loop by investigating the overlap between the Plan and Execute phases, possibly running several P-E cycles in parallel, under explicit consistency and conflict-resolution guarantees. More precisely, the objectives of the thesis is to design, formalize, and evaluate a concurrent coordination pattern for MAPE-K, referred to as MACPE-K, for service orchestration in the Cloud-Edge-IoT continuum. The work will formally establish the conditions under which the Plan and Execute phases can overlap while guaranteeing consistency and resolving action-level conflicts, and then model the end-to-end responsiveness/cost trade-off, including the Monitoring and Analysis phases. The dynamic tuning of the degree of concurrency and of the triggering policies will be investigated in order to preserve responsiveness under varying workloads and environmental conditions. Service placement will serve as the primary case study, but a second self-management task (e.g., elastic autoscaling or fault recovery) will also be instantiated to substantiate the genericity of the pattern, covering both stateless and stateful services
Principales activités
The work of the PhD will be organized in several phases:
- State of the art on MAPE-K coordination patterns (single loop, hierarchical, decentralized, peer-to-peer, etc.), parallel orchestration tools (Concerto, Kubernetes, etc.), and concurrency control applied to self-adaptive systems.
- Mathematical formalization of the concurrent pattern (queueing theory, timed automata, etc.) including Monitoring and Analysis costs, specification of an action-level conflict detection and resolution policy together with a correctness proof, and instantiation on at least two case studies covering both stateless and stateful services.
- Implementation of the proposed pattern in a simulator and in a prototype based on existing IaC tools, with experimental evaluation and overhead measurements on the experiment-driven platform such as Grid'5000 or SLICES-RI .
Compétences
We are looking for a candidate holding (or about to obtain) a Master's degree or engineering diploma in computer science or a related field, with:
- Strong interest in research questions related to autonomic systems, performance modeling, or adaptive algorithms.
- Good understanding of distributed systems, cloud/edge computing, and containerized architectures (e.g., Docker, Kubernetes).
- Solid programming abilities (Python), sufficient to build prototypes supporting scientific exploration.
- Experience with experimental design, benchmarking, or evaluation methodologies is a plus.
- Curiosity, rigour, autonomy, and a taste for research.
- Very good communication skills in oral and written English
Avantages
- Subsidized meals
- Partial reimbursement of public transport costs
- Leave: 7 weeks of annual leave + 10 extra days off due to RTT (statutory reduction in working hours) + possibility of exceptional leave (sick children, moving home, etc.)
- Possibility of teleworking (90 days / year) and flexible organization of working hours
- Social, cultural and sports events and activities
- Access to vocational training
- Social security coverage
Rémunération
2300 euros gross salary /month
