Systems of Twinned Systems: A Systematic Literature Review

This paper presents a systematic literature review of 80 studies on "systems of twinned systems," a paradigm that integrates the System of Systems and Digital Twin concepts to address modern engineering complexity, resulting in a new classification framework compatible with existing theories.

The Gist

Imagine you are the conductor of a massive, chaotic orchestra. But instead of just musicians, your orchestra is made up of thousands of different instruments, robots, cars, and even people, all scattered across the globe. Some play jazz, some play classical, and some are still learning their notes.

This is the challenge of modern engineering: Systems are getting too big and too complex to manage alone.

This paper is a "systematic review," which is like a giant detective story where the authors scoured through over 2,500 research papers to find the best 80 that answer a specific question: How do we combine two powerful ideas—Digital Twins and Systems of Systems—to manage this chaos?

Here is the breakdown in simple terms, using some fun analogies.

1. The Two Main Characters

To understand the paper, you need to know the two main characters in this story:

• The System of Systems (SoS): Think of this as a giant potluck dinner. Everyone brings their own dish (a car, a traffic light, a power plant). They are all independent; the car owner decides when to eat, and the power plant decides when to cook. But they all come together to feed the community. The challenge is getting them to work together without a single boss telling everyone exactly what to do.
• The Digital Twin (DT): Think of this as a perfect, magical mirror. If you have a real physical car, its Digital Twin is a virtual copy in a computer that knows everything about the real car. If the real car gets a flat tire, the mirror shows it instantly. If you push a button in the mirror, the real car reacts. It's a two-way conversation between the real world and the digital world.

2. The New Idea: "Systems of Twinned Systems" (SoTS)

The authors realized that we are starting to mix these two concepts. We aren't just managing a potluck of real cars anymore; we are managing a potluck where every single dish has its own magical mirror.

They call this a System of Twinned Systems (SoTS).

• The Analogy: Imagine a smart city. Instead of just watching traffic cameras, every car, every traffic light, and every pedestrian has a "Digital Twin" in the cloud. These twins talk to each other. If a car's twin sees a red light ahead, it tells the car's twin to slow down. If a power plant's twin sees a storm coming, it tells the city's grid twin to reroute energy.
• The Goal: By having these twins talk to each other, we can solve complex problems (like traffic jams or energy shortages) much faster and smarter than if we just looked at the real things.

3. How Do They Talk? (The Architectures)

The paper found that these "Twin Potlucks" are organized in different ways, kind of like different ways to run a team:

• The "Boss" Style (Directed SoTS): One central Digital Twin acts as the General Manager. It tells all the other twins what to do. Example: A central city computer telling all traffic lights to turn green at the same time to clear a jam.
• The "Negotiator" Style (Acknowledged SoTS): There is a central manager, but the individual twins have their own goals. They negotiate. Example: A power grid manager asks a factory twin to use less power, but the factory twin says, "I can only do that if I get a discount on my electricity bill."
• The "Friend Group" Style (Collaborative SoTS): There is no boss. The twins just hang out and help each other because they agreed to a common goal. Example: A group of self-driving cars deciding on their own to form a "platoon" to save fuel, without a central computer forcing them.
• The "Wild Card" Style (Virtual SoTS): The twins are so independent they just show up when they want and do their own thing, and the system adapts on the fly. Example: Emergency drones arriving at a disaster zone and figuring out a rescue plan together as they go.

4. What Did the Authors Find? (The Results)

After reading 80 deep-dive studies, here is the "state of the union":

• It's Mostly About Manufacturing: Most of these "Twin Potlucks" are happening in factories. They are using it to make assembly lines run smoother.
• The "Boss" is Still Popular: Most systems still rely on a central controller (Directed or Acknowledged). The "Friend Group" style where everyone is totally free is still rare and hard to build.
• The Twins are Mostly "Watchers": Most Digital Twins are great at watching and simulating (predicting what will happen), but they aren't great at acting on their own yet. They are like a very smart advisor, not a decision-maker.
• Security is a Blind Spot: While everyone is building these cool systems, very few are actually testing if they are secure. It's like building a high-tech bank vault but forgetting to check if the door locks properly.
• It's Still a Baby: The technology is mostly in the "Prototype" phase. We have built cool models and tested them in labs, but very few are running in the real world 24/7.

5. The Big Problems (The "But...")

The authors point out three major hurdles standing in the way of a perfect "Twin Potluck":

1. Speaking Different Languages: One factory's Digital Twin might speak "Python," while a power plant's twin speaks "Java." They can't understand each other. We need a universal translator (Standards).
2. The "Surprise" Factor: In a System of Systems, weird things happen that you didn't plan for (Emergence). If 1,000 cars all decide to take a shortcut at the same time, it causes a new kind of jam. We don't have good tools to predict these surprises yet.
3. Lack of Rules: There aren't enough official rules or blueprints (Architectures) on how to build these things. Everyone is reinventing the wheel.

The Bottom Line

This paper is a wake-up call. We have the tools to build a world where everything is connected and "mirrored" in the digital realm. It's a powerful idea that could fix traffic, save energy, and make factories run like clockwork.

However, we are currently in the "Wild West" phase. We have the horses and the wagons, but we don't have the roads, the traffic laws, or the map yet. The authors are asking researchers and engineers to stop just building cool prototypes and start building the rules, the standards, and the safety checks so this technology can actually work in the real world.

Technical Summary

Here is a detailed technical summary of the paper "Systems of Twinned Systems: A Systematic Literature Review" by Adesanya et al.

1. Problem Statement

Modern engineered systems are facing unprecedented complexity due to scale, interconnectivity, and the heterogeneity of digital and physical components. Two distinct paradigms have emerged to address specific aspects of this complexity:

• System of Systems (SoS): Focuses on aggregating independent, distributed subsystems to achieve higher-level goals while maintaining operational and managerial independence. It emphasizes flexibility and scalability but traditionally treats constituents as "black boxes" with weak coupling.
• Digital Twins (DT): Focuses on the tight, bidirectional coupling between a physical entity and its digital representation, enabling real-time monitoring, simulation, and control.

The Gap: While both paradigms are mature individually, their convergence creates a new class of systems called Systems of Twinned Systems (SoTS). In an SoTS, multiple digital twins (and their physical counterparts) are organized according to SoS principles. However, the field lacks a unified understanding of how these systems are architected, what their technical characteristics are, and what challenges hinder their maturity. There is no comprehensive state-of-the-art review defining the landscape of SoTS.

2. Methodology

The authors conducted a Systematic Literature Review (SLR) following rigorous empirical software engineering guidelines (Kitchenham et al., Petersen et al.).

• Search Strategy:
  • Databases: Scopus, Web of Science, ACM Digital Library, and IEEE Xplore.
  • Search String: Focused on combinations of "digital twin" and "system of systems" (including synonyms like "aggregated digital twin," "system of twinned systems").
  • Scope: Peer-reviewed literature only; grey literature excluded.
• Selection Process:
  • Initial Screening: Retrieved 317 studies; removed duplicates to get 196.
  • Exclusion Criteria: Applied criteria regarding accessibility, peer-review status, and relevance (must discuss both DT and SoS).
  • Quality Assessment: Used a checklist based on research questions (clarity of SoS/DT description, tangible contributions).
  • Snowballing: Performed forward and backward snowballing on the initial set to ensure saturation.
• Final Corpus: 80 primary studies were included after screening over 2,500 potential papers.
• Data Analysis: Combined content analysis (categorization) and narrative synthesis. The authors developed a custom classification framework to analyze the data vertically (within categories) and horizontally (across categories).

3. Key Contributions

The paper makes three primary contributions to the field:

1. Definition of SoTS: The authors formally define a System of Twinned Systems as "digitally twinned systems organized by system-of-systems (SoS) principles, in which digitally twinned systems may act as autonomous constituents and collaborate to achieve complex goals."
2. Classification Framework: They propose a taxonomy for SoTS that is backward compatible with existing SoS and DT theories. The framework includes:
   • Paradigms & Characteristics: (e.g., Autonomy, Independence, Distribution, Emergence).
   • Architectural Patterns: Six distinct patterns derived from the study:
     • Directed SoTS: Central DT controller orchestrates constituents.
     • Acknowledged SoTS: Central DT coordinates, but constituents negotiate goals.
     • Collaborative SoTS: No central controller; coordination via choreography (peer-to-peer).
     • Virtual SoTS: Highly autonomous constituents with on-the-fly goal negotiation.
     • Specialized DTs: Loosely coordinated DTs for the same constituent.
     • Specialized DTs & Systems: Overlapping sets of specialized DTs and systems.
   • Means of Composition: Mechanisms like orchestration vs. choreography.
3. Comprehensive State-of-the-Art Analysis: The review answers seven specific Research Questions (RQs) covering motivations, architectures, technical characteristics, non-functional properties, maturity, and implementation technologies.

4. Key Results & Findings

RQ1: Motivations and Domains

• Motivations: The primary drivers are Optimization (37.5%), Integration (31.3%), Validation (18.8%), and Maintainability (12.5%).
• Domains: Manufacturing is the dominant domain (40%), followed by Automotive (11.3%) and Smart Cities (7.5%).
• Intent: Most studies (61.3%) focus on combining multiple DTs into an SoS, rather than twinning an existing SoS as a whole.

RQ2: Architectures and Constituents

• Architecture: Acknowledged SoTS (38.8%) and Directed SoTS (32.5%) are the most common. Highly dynamic architectures like Collaborative (23.8%) and Virtual (5.0%) are less frequent, suggesting a reliance on centralized control.
• Constituents: The vast majority (77.5%) focus on Physical Systems. Cyber-Physical Systems (11.3%) and Cyber-Physical-Human Systems (8.8%) are less common.

RQ3 & RQ4: Technical Characteristics

• DT Autonomy: Most DTs are fully autonomous (82.5%). Digital Shadows (passive) and Human-Supervised models are rare.
• Services: The most common services are Real-time monitoring (98.8%), Simulation (96.3%), and Optimization (85.0%).
• SoS Dimensions: Studies heavily address Distribution (92.5%) and Independence (88.8%). However, dynamic aspects like Evolution (37.5%) and Reconfiguration (43.8%) are under-addressed.
• Emergence: Weak emergence (37.5%) is the most reported type. Strong emergence is rare (7.5%), and 35% of studies do not address emergence at all.

RQ5: Non-Functional Properties (NFPs)

• Reliability: Addressed architecturally in 51.3% of studies (e.g., fallback mechanisms), but rarely formally modeled (2.5%) or validated (3.8%).
• Security: Addressed architecturally in 23.8% of studies. Explicit modeling and validation are extremely low (<4%).

RQ6: Maturity

• Technology Readiness Level (TRL): The field is in an early-to-mid stage.
  • Demo Prototypes (43.8%) and Proof-of-Concept (20.0%) dominate.
  • Only 1.3% of studies report fully Operational systems.
• Evaluation: 90% of studies use Validation (prototyping/simulation), while only 10% use Evaluation (industrial case studies/action research).
• Standards: Less than half (45%) of studies use standards. When used, they are often DT-specific (e.g., OPC UA, AAS) rather than SoS-specific.

RQ7: Implementation Technologies

• Languages: Python (27.5%) and Java (17.5%) are the most used.
• Tools: Heavy reliance on modeling/simulation tools (MATLAB, Simulink, Gazebo) and data management (MongoDB).
• Frameworks: Eclipse Ditto, ROS, and cloud tools (Docker, Azure) are common.

5. Significance and Future Directions

The paper highlights that SoTS is an emerging field with significant gaps between theoretical potential and engineering reality.

• Architectural Deficit: There is a lack of established, flexible architectures for SoTS. The dominance of "Directed" and "Acknowledged" patterns suggests current research struggles to support the dynamic, decentralized nature required for true SoS.
• Standardization Gap: The lack of unified standards for SoS composition hinders interoperability, especially in critical domains like automotive and smart cities. The authors anticipate upcoming ISO 23247 extensions (Digital Thread and Composition) to address this.
• Maturity Lag: The field is stuck in the "prototype" phase with limited real-world validation. There is a critical need for more industrial case studies and empirical evaluations.
• Emergent Behavior: Current SoTS designs often ignore or fail to manage emergent behaviors. The authors suggest leveraging DT capabilities (e.g., active experimentation, faster-than-real-time simulation) to better predict and manage emergence.

Conclusion: The authors call for the development of reference architectures, technical standards, and empirical case studies to mature the field of Systems of Twinned Systems, moving from theoretical concepts to robust, operational cyber-physical ecosystems.