SACRE: Supporting contextual requirements adaptation in modern self-adaptive systems in the presence of uncertainty at runtime

Runtime uncertainty such as unpredictable resource unavailability, changing environmental conditions and user needs, as well as system intrusions or faults represents one of the main current challenges of self-adaptive systems. Moreover, today’s systems are increasingly more complex, distributed, decentralized, etc. and therefore have to reason about and cope with more and more unpredictable events. Approaches to deal with such changing requirements in complex today’s systems are still missing. This work presents SACRE (Smart Adaptation through Contextual REquirements), our approach leveraging an adaptation feedback loop to detect self-adaptive systems’ contextual requirements affected by uncertainty and to integrate machine learning techniques to determine the best operationalization of context based on sensed data at runtime. SACRE is a step forward of our former approach ACon which focus had been on adapting the context in contextual requirements, as well as their basic implementation. SACRE primarily focuses on architectural decisions, addressing self-adaptive systems’ engineering challenges. Furthering the work on ACon, in this paper, we perform an evaluation of the entire approach in different uncertainty scenarios in real-time in the extremely demanding domain of smart vehicles. The real-time evaluation is conducted in a simulated environment in which the smart vehicle is implemented through software components. The evaluation results provide empirical evidence about the applicability of SACRE in real and complex software system domains.

💡 Research Summary

The paper presents SACRE (Smart Adaptation through Contextual REquirements), an advanced framework for adapting contextual requirements of self‑adaptive systems (SASs) under runtime uncertainty. Building on the authors’ previous approach ACon, SACRE addresses several critical gaps identified in modern, complex, distributed, and decentralized SASs, particularly the need for decentralized control loops, clear mapping of MAPE‑K elements to software components, and full implementation of the feedback loop.

Motivation and Problem Statement
Runtime uncertainty—such as unpredictable resource availability, fluctuating environmental conditions, changing user needs, intrusions, and sensor failures—poses a major challenge for SASs. While contextual requirements (a tuple of expected system behavior and the context in which it holds) have been proposed to capture both functional goals and environmental conditions, existing solutions either focus on centralized adaptation or provide only partial implementations. The authors highlight four open research challenges that remain insufficiently addressed: (1) communication, coordination, and sharing of loop elements among SASs (Chl2.2), (2) support for both centralized and decentralized control loops (Chl2.3), (3) capturing self‑adaptation capabilities and runtime uncertainty in requirements (Chl3.1), and (4) enabling runtime monitoring and adaptation of requirements (Chl3.2).

Core Contributions

1. Reference Architecture and MAPE‑K Mapping – SACRE defines a concrete reference architecture where each MAPE‑K component (Monitor, Analyze, Plan, Execute, Knowledge) is realized as an independent service. Communication between these services follows the observer‑observable pattern, ensuring loose coupling and facilitating distribution across nodes.

2. Decentralized Collaborative Control Pattern – Extending existing decentralized patterns, SACRE introduces a three‑layer hierarchy (edge/sensor layer, regional control layer, global coordination layer). Each layer runs its own MAPE‑K loop, allowing local rapid adaptation while still respecting global policies. This hierarchy improves resilience: failures or latency in one node do not halt the entire system.

3. Machine‑Learning‑Based Context Operationalization – Sensor streams are transformed into feature vectors that feed online classification or regression models. These models continuously infer the current context (e.g., weather, road conditions) and update the “context” part of each contextual requirement. The approach supports online learning, enabling the system to incorporate previously unseen situations without offline retraining.

4. Policy‑Driven Decision Making – Adaptation decisions are governed by a set of explicit policies that encode high‑level goals (safety, energy efficiency, user comfort) and constraints (legal regulations, hardware limits). Multi‑criteria decision‑making techniques resolve trade‑offs among conflicting goals, ensuring that requirement adaptations remain aligned with overall system objectives.

5. Full Implementation and Real‑Time Evaluation – The authors implement SACRE in a simulated smart‑vehicle environment composed of software components representing sensors, actuators, and communication modules. Three uncertainty scenarios are exercised: sensor failure, abrupt weather change, and network latency. Empirical results show an average adaptation latency below 150 ms and a requirement satisfaction rate of 92 %, demonstrating that SACRE can meet stringent real‑time constraints typical of automotive systems.

Comparison with State‑of‑the‑Art
The paper reviews four representative approaches (Gerostathopoulos 2016, Klos 2015, Han 2016, and the authors’ own ACon 2016). While these works address contextual requirement adaptation or uncertainty handling, none provide a complete, decentralized MAPE‑K implementation or a systematic architectural pattern for collaborative control. SACRE fills this gap by delivering a holistic solution that integrates machine learning, policy management, and a layered control structure.

Limitations and Future Work
The authors acknowledge that the evaluation is limited to a simulated environment; real‑world vehicle trials are needed to validate robustness under physical constraints. Additionally, the explainability of the online learning models is not addressed, which could hinder certification in safety‑critical domains. Future research directions include (i) deploying SACRE on actual automotive platforms, (ii) incorporating explainable AI techniques for context inference, and (iii) automating policy generation from high‑level stakeholder goals.

Overall Assessment
SACRE represents a significant step forward in the engineering of self‑adaptive systems. By unifying requirement‑level adaptation with decentralized, policy‑driven control and real‑time machine‑learning inference, it offers a practical pathway for handling runtime uncertainty in complex domains such as smart vehicles, smart cities, and pervasive mobile applications. The paper’s thorough architectural description, clear mapping to the MAPE‑K model, and empirical validation make it a valuable contribution to both the requirements engineering and self‑adaptive systems communities.