A Knowledge-Centric Communication For Autonomous Cislunar Networks

arXiv:2608.04817 · eess.SP · Submitted 2026-08-05 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Electrical Engineering. The day in electrical engineering: signals, systems and the hardware that carries them.

Rosa: I'm Rosa.

Dev: And I'm Dev, here with Taro.

Rosa: We're looking at the paper: "A Knowledge-Centric Communication For Autonomous Cislunar Networks".

Dev: Here's the gist of the paper. Future lunar infrastructure requires communication as a persistent service rather than a mission-specific capability This work introduces a knowledge-centric communication framework in which a digital twin continuously integrates delayed observations, communication physics, learned models, uncertainty quantification, and mission objectives into an evolving Operational Knowledge State <ref:2608.04817#pg6> The Core Problem Addressed The central motivation stems from the breakdown of the assumption that autonomous decisions can be supported by sufficiently accurate operational information when communication is delayed and intermittent in cislunar space <ref:2608.04817#pg4> This structural unavailability means that autonomous communication decisions must integrate delayed observations, communication physics, predictive models, mission objectives, and quantified uncertainty into what the infrastructure currently knows and does not know about itself and its environment <ref:2608.04817#pg4> The framework addresses this by defining two key metrics: Knowledge Entropy, which quantifies how much remains unknown as the differential entropy of the posterior belief <ref:2608.04817#pg4>, and Knowledge Freshness, an Age of Information (AoI)-derived clock associated with the most recently assimilated evidence <ref:2608.04817#pg4> The Operational Knowledge State Construction The Operational Knowledge State is defined as a time-varying graph K(t) = G(t), and the set of posterior beliefs for active links, where each belief bij(t) = p(sij (t) O1:t) <ref:2608.04817#pg9> This state is constructed by continuously fusing four heterogeneous streams: delayed observations describe a past rather than a present state <ref:2608.04817#pg10>, communication physics constrains what that state can plausibly be independent of any single observation <ref:2608.04817#pg10>, learned models capture the residual behavior that physics cannot explain together with their own predictive uncertainty <ref:2608.04817#pg10>, and mission objectives specify how precisely each part of the environment must be known <ref:2608.04817#pg10> Consistency Properties of the Framework The proposed inference mechanism specializes established Gaussian fusion and covariance propagation results to the KCDT, stating that informative evidence should reduce posterior uncertainty, whereas uncertainty should not decrease during an observation gap without new evidence <ref:2608.04817#pg10> Furthermore, Knowledge Entropy is distinct from AoII because it quantifies posterior uncertainty when the true state is not promptly observable <ref:2608.04817#pg10> The framework also establishes consistency conditions for standard Gaussian information-form fusion and covariance prediction as the inference layer <ref:2608.04817#pg10> Knowledge-to-Decision Interface The Operational Knowledge State becomes actionable through an admissibility-and-risk interface defined by the set of admissible links Eadm(t) where uncertainty and evidence age satisfy mission requirements <ref:2608.04817#pg11> For every admissible link, a normalized knowledge-aware cost cij(t) is defined as a function of propagation delay, entropy, freshness, and SNR <ref:2608.04817#pg14> This leads to concrete reference policies for adaptive routing, proactive link switching, and uncertainty-aware resource allocation <ref:2608.04817#pg14> Validation Against Flight Data The framework was evaluated using Monte Carlo simulations under the environment and protocol described in Section E, and a physically grounded comparison based on Longjiang-2 flight data <ref:2608.04817#pg18> The results show that distributed observations produced more informative Operational Knowledge than any individual communication source Furthermore, the knowledge-centric belief increased by 0.176 nats during the anomaly and remained elevated briefly afterward, which is consistent with reports that the jump corrupted some, but not all, of the affected packets <ref:2608.04817#pg19> The framework suggests a shift from state-centric communication to knowledge-centric reasoning realized through the KCDT <ref:2608.04817#pg4> The framework is a starting point rather than a complete solution, and the shift from state-centric to knowledge-centric reasoning is expected to apply beyond Earth to Moon communications where autonomous systems must act on what they currently know rather than on what may eventually become known <ref:2608.04817#pg22> The paper opens a research direction in which the representation, quantification, and lifecycle of operational knowledge become as central to communication theory as the traditional concept of the bit <ref:2608.04817#pg22> References

1: National Aeronautics and Space Administration, “Moon to Mars architecture: Lunar communications, positioning, navigation, and timing (CPNT) architecture,” NASA, Washington, DC, USA, Tech.

Rosa: Rep., 2023

2: S.

Dev: G.

Rosa: C¸ etin, B.

Title and authors: Rosa: So we're looking at this paper, "A Knowledge-Centric Communication For Autonomous Cislunar Networks." It’s about how future lunar infrastructure needs communication to be more of a constant service instead of just something you do for one specific mission.

Dev: Right, so the core issue they are tackling is that we can't rely on having perfect, up-to-date information when communication is delayed and intermittent in space. The assumption that you can make autonomous decisions based on good operational data breaks down there because of the propagation delay and visibility issues in cislunar space.

Taro: That sounds like the fundamental problem they are addressing, where you need to integrate everything—the delays, the physics of communication, learned models, uncertainty quantification—into what you know and what you don't know about the environment.

Rosa: Exactly. They propose this knowledge-centric framework where a digital twin constantly fuses all that stuff together into what they call an evolving operational knowledge state.

Dev: It’s pretty dense. They define two key metrics to measure how good that knowledge is, which are Knowledge Entropy, measuring the uncertainty of what remains unknown, and Knowledge Freshness, which is an age of information clock tied to the newest evidence assimilated.

Taro: So it’s not just about having data; it’s about quantifying how much you *don't* know and how old the information is before you make a move.

Rosa: That brings us into how they build this operational knowledge state, which they define as a time-varying graph, or K of t equals G of t.

Dev: This state is built by continuously fusing four different streams: delayed observations that describe the past instead of the present, communication physics that constrain what that past state could possibly be regardless of any single observation, learned models capturing behavior physics alone can't explain along with their own uncertainty, and mission objectives defining exactly how precise you need to be about everything.

Taro: It sounds like they’re trying to get a holistic picture by combining all those different sources of information.

Rosa: And this framework has some consistency properties too. They show that informative evidence should reduce the posterior uncertainty, but that uncertainty shouldn't decrease during an observation gap without new evidence coming in.

Dev: They also distinguish Knowledge Entropy from Age of Information because one measures posterior uncertainty when the true state isn't immediately observable, which is a key difference for this application.

Taro: It’s interesting how they establish consistency conditions for standard Gaussian information-form fusion and covariance prediction as the layer for their inference mechanism.

Rosa: So once they have this knowledge state, it becomes actionable through an admissibility and risk interface defined by admissible links, E adm of t.

Dev: For every link that is admissible, they define a normalized knowledge aware cost, c of t, which depends on propagation delay, entropy, freshness, and the signal to noise ratio.

Taro: That cost function should guide things like adaptive routing or proactive link switching based on those metrics rather than just instantaneous measurements.

Rosa: And this leads to concrete reference policies for things like adaptive routing and uncertainty aware resource allocation.

Dev: The validation involved Monte Carlo simulations under a specific environment, and they also did a comparison based on Longjiang two flight data. They found that distributed observations actually produced more informative operational knowledge than any single communication source.

Taro: That’s significant because it means relying on one spot isn't enough; spreading out the observations helps build the knowledge base better.

Rosa: And during an anomaly in their testing, the knowledge-centric belief actually increased by zero point one hundred seventy-six nats, which they link to reports that some packets were corrupted but not all of them.

Dev: That suggests that even when things go wrong, this framework helps capture the change in understanding better than a traditional system might.

Taro: It’s a nice result because it shows the knowledge itself is evolving and tracking those anomalies, rather than just being stuck on old data points.

Rosa: So, to wrap up on the improvements they suggest, they are moving from state-centric communication to knowledge-centric reasoning realized through this digital twin concept.

Dev: They argue this shift is expected to apply beyond Earth to Moon communications where autonomous systems need to act based on what they currently know rather than waiting for something that might eventually become known.

Taro: What I find interesting is the paper opens up a direction where the representation and lifecycle of operational knowledge become as central to communication theory as the traditional concept of a bit.

Rosa: It really suggests that for autonomous systems out there, we need to focus on what you know right now instead of waiting for future perfect information.

Dev: This whole knowledge-centric communication framework, introduced in "A Knowledge-Centric Communication For Autonomous Cislunar Networks," is a starting point rather than a finished solution.

Taro: It’s certainly a complex area, and I think we need to keep watching how they develop this representation of operational knowledge.

Rosa: We've seen how this framework uses metrics like Knowledge Entropy and Freshness to guide decisions based on the current state of our understanding.

The paper's summary: Rosa: So we’re looking at how these authors summarize their main idea for this paper, which is all about shifting communication from just sending data to building a whole knowledge base for autonomous systems in space.

Dev: Right, so they’re talking about moving away from just thinking about the current state of the system and starting to think about what you don't know yet, and how old that information is.

Rosa: Exactly. They’re proposing this digital twin concept—the Knowledge-Centric Digital Twin—where all these things like delayed observations, the physical laws of communication, learned models, and even mission goals get fused together into one evolving operational knowledge state.

Dev: It sounds like they define two specific metrics to track that knowledge base: Knowledge Entropy and Knowledge Freshness.

Rosa: That’s right. Knowledge Entropy measures how much is still unknown as the differential entropy of the posterior belief, which tells you exactly how uncertain you are about what's going on out there.

Dev: And Knowledge Freshness is an age of information clock, basically measuring how far behind your most recent piece of data is from the present moment.

Rosa: So instead of just having a static picture of the environment, they’re building a dynamic picture that constantly updates based on new evidence and how old that evidence is.

Dev: It means when an autonomous system needs to make a decision, it doesn't just look at the current state; it looks at this whole knowledge base and asks what its uncertainty and information age are.

Rosa: And they create an interface for this, called the admissibility-and-risk interface, where they define admissible links based on whether their uncertainty and evidence age meet mission requirements.

Dev: For every link that’s admissible, they calculate a knowledge aware cost that depends on the delay of propagation, the entropy level, how fresh it is, and the signal to noise ratio.

Rosa: That cost function is what drives things like proactive link switching or uncertainty aware resource allocation instead of just sticking to whatever you have right now.

Dev: The validation they did using simulations and flight data showed that having distributed observations actually gave them more useful knowledge than any single communication source could provide.

Rosa: And during their testing, the system’s knowledge level even increased when things went a little weird, which they suggest is consistent with reports that some of the packets were corrupted but not all of them.

Dev: So it’s about making autonomous decisions based on what you currently know, rather than waiting for something that might eventually become known.

Rosa: This whole framework suggests a real shift away from state-centric communication toward knowledge-centric reasoning in space networks.

The paper's improvements: Rosa: So we’re looking at what these authors suggest as improvements for this knowledge-centric communication framework for lunar networks. It’s about how they make it actually work in a real, messy environment out there.

Dev: Right, they aren't just giving us a theory; they are defining the operational steps—the policies—for how the system should behave when it encounters uncertainty and information decay.

Rosa: They show that this knowledge-centric reasoning allows for proactive link switching and uncertainty aware resource allocation, which is way better than just reacting to what happens on a single channel.

Dev: They define this through an admissible links set, meaning the system only considers communication paths where the uncertainty and age of evidence meet specific mission requirements.

Rosa: That means if the knowledge base gets too uncertain or too old for a link, it gets cut from consideration automatically.

Dev: And they also address that issue of uncertainty not decreasing during an observation gap, which is a common problem in real-time systems where you just wait for more data to show up.

Rosa: They establish consistency conditions for standard Gaussian fusion and covariance prediction as the inference layer, which gives engineers a solid mathematical foundation for how to actually build the digital twin.

Dev: It’s important because it means they are taking established fusion results and applying them specifically to this kind of delayed, intermittent space communication.

Rosa: And they also show that their knowledge entropy metric is distinct from the age of information clock because one specifically measures uncertainty when the true state isn't immediately observable.

Dev: So it’s a more precise way to measure how much you genuinely don't know versus just how old your last piece of data is.

Rosa: It opens up a future where autonomous systems can make decisions based on what they currently know, instead of waiting for perfect information that might never arrive in time.

Dev: This whole shift from state-centric communication to knowledge-centric reasoning is something that should apply beyond just Earth and Moon communications, to any system where autonomy is key.

Conclusion: Rosa: So we’re wrapping up on this paper, "A Knowledge-Centric Communication For Autonomous Cislunar Networks." Basically, they are proposing a way for space infrastructure to communicate not just by sending bits, but by building a whole knowledge base that evolves over time.

Dev: Right, so the main implication is shifting from systems that react to immediate signals to systems that act on what they currently know about their environment.

Rosa: They show how this knowledge-centric approach lets autonomous networks make better choices, like choosing the right communication link or allocating resources in a way that accounts for how old the information is.

Dev: It’s about making those decisions based on an admissibility-and-risk interface, which means only using links that meet strict requirements for uncertainty and age of evidence.

Rosa: And they provide concrete policies for things like adaptive routing and resource allocation based on that knowledge state, instead of just guesswork.

Dev: The validation with flight data supports this because distributed observations actually gave them more useful knowledge than any single point could provide, which is a big deal for reliability.

Rosa: It also shows that this framework can handle anomalies better because the knowledge base itself updates to reflect what’s happening, even when some packets get corrupted.

Dev: The core idea is that this shift toward reasoning based on current knowledge, rather than waiting for future perfect information, should apply wherever autonomous systems operate in space.

Rosa: It really suggests that for future lunar or deep space infrastructure, the representation and management of operational knowledge need to be just as important as the traditional concept of a bit.

Dev: It’s a starting point with some caveats about how complex this digital twin becomes, but it sets a new direction for how we design resilient networks.

Taro: I think what really matters is that this moves the focus from just transmitting data to actually understanding and managing the information itself in real time.

Rosa: Exactly. It’s about having a system that knows what it knows and when it doesn't, which makes recovery much smoother when things go wrong out there.

Dev: We’re ready for the next paper, so we’ll keep an eye out for more work on how AI can manage these complex knowledge states in real-time control loops.

Interdisciplinary Research Center for Communication Systems and Sensing at King Fahd University of Petroleum and Minerals

eess.SP

Submitted: 2026-08-05

Updated: 2026-11-09

Comments: 23 pages, 8 Figures, 2 Tables (Accepted)

Journal ref: IEEE OJCOMS 2026

Code: https://github.com/afanali85-tech/Knowledge-Centric-Digital-Twin-KCDT-

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 74/100

The gist: Future lunar infrastructure requires communication as a persistent service rather than a mission-specific capability This work introduces a knowledge-centric communication framework in which a…

Key concepts

Operational Knowledge State
This is a time-varying graph representing everything the lunar infrastructure knows about itself and its environment. It's built by continuously fusing delayed observations, communication physics constraints, learned models, and mission objectives to form a complete picture.
Knowledge Entropy
This metric measures how much information remains unknown in the system. It quantifies the uncertainty of the current belief state as if the true state were not immediately observable. High entropy means high uncertainty.
Knowledge Freshness
This is an Age of Information (AoI)-derived clock that tracks when each piece of evidence was last assimilated. It tells autonomous systems how old their information is, helping them avoid relying on stale data during communication gaps.

Terminology

Summary

Future lunar infrastructure requires communication as a persistent service rather than a mission-specific capability This work introduces a knowledge-centric communication framework in which a digital twin continuously integrates delayed observations, communication physics, learned models, uncertainty quantification, and mission objectives into an evolving Operational Knowledge State <ref:2608. p. 0

Where the paper says it

Cislunar communication violates this assumption because propagation delay is fundamentally limited by the speed of light, visibility is governed by orbital geometry, and autonomous decisions often precede confirmatory observations. Here we introduce a knowledge-centric communication framework in which a digital twin continuously integrates delayed observations, communication physics, learned models, uncertainty quantification, and mission objectives into an evolving Operational Knowledge State.

04817#pg6>

The Core Problem Addressed

The central motivation stems from the breakdown of the assumption that autonomous decisions can be supported by sufficiently accurate operational information when communication is delayed and intermittent in cislunar space <ref:2608.04817#pg4> This structural unavailability means that autonomous communication decisions must integrate delayed observations, communication physics, predictive models, mission objectives, and quantified uncertainty into what the infrastructure currently knows and does not know about itself and its environment <ref:2608. p. 1

Where the paper says it

: A Knowledge-Centric Communication For Autonomous Cislunar Networks: IEEE Open Journal of the Communications Society COMMUNICATION 1948 1990s – 2010s 2020s this work NEW Classical Communication Network-Centric Communication Semantic Communication Knowledge-Centric Communication faithful symbol reproduction metric: bit error rate reliable delivery under delay & disruption metric: throughput, AoI preservation of meaning & task relevance metric: semantic similarity reasoning over what is known, and not known metric: entropy, freshness DIGITAL TWIN ±σ Virtual Counterpart Predictive & Bidirectional AI-Native, Distributed Agentic / Cognitive Twin mirrors a physical asset across its lifecycle hybrid physics–data models, continual updating, UQ & V&V knowledge-graph-based architectures for cyber–physical systems twin contributes to the decision loop, not only feeds it ONBOARD AI Onboard Compression & Preprocessing Distributed Edge Intelligence Foundation Models, Space Cloud onboard processing limited to data compression collaborative sensing, federated learning, autonomous inference onboard foundation models embedded in the comms infrastructure communication frameworks optimize channels, routing, coding, and resource allocation • digital twins represent, synchronize, and predict physical systems • onboard intelligence extracts actionable information from sensed data a shared assumption, rarely stated autonomous decisions can be supported by sufficiently accurate operational information – i. , a subsystem can eventually reconcile its internal state against ground truth breaks down Earth – Moon infrastructure breaks this assumption • Propagation delay is bounded below by the physics of light travel time • Intermittent visibility is set by orbital geometry, not by deployment choice • Decisions must often execute before any corroborating observation can arrive manufacturing, aerospace-sustainment, and Earth-system twins can treat synchronization delay as an engineering nuisance, eventually reconciled against ground truth – in cislunar space that reconciliation is structurally unavailable, not a modelling simplification Autonomous communication decisions must instead integrate delayed observations, communication physics, predictive models, mission objectives, and quantified uncertainty into what the infrastructure currently knows, and does not know, about itself and its environment (this work) FIGURE 1: Three research threads converging on autonomy.

04817#pg4> The framework addresses this by defining two key metrics: Knowledge Entropy, which quantifies how much remains unknown as the differential entropy of the posterior belief <ref:2608. p. 2
Where the paper says it

We define Knowledge Entropy as the entropy of the approximate posterior distribution that the twin maintains over the unobserved components of infrastructure state. It quantifies how much remains unknown.

04817#pg4>, and Knowledge Freshness, an Age of Information (AoI)-derived clock associated with the most recently assimilated evidence <ref:2608. p. 0
Where the paper says it

Here we introduce a knowledge-centric communication framework in which a digital twin continuously integrates delayed observations, communication physics, learned models, uncertainty quantification, and mission objectives into an evolving Operational Knowledge State. To quantify what is known, how uncertain it remains, and how current it is, we define two quantities over the fused Bayesian belief maintained for each communication link: Knowledge Entropy, the differential entropy of the posterior belief, and Knowledge Freshness, an Age of Information (AoI)-derived clock associated with the most recently assimilated evidence.

04817#pg4>

The Operational Knowledge State Construction

The Operational Knowledge State is defined as a time-varying graph K(t) = G(t), and the set of posterior beliefs for active links, where each belief bij(t) = p(sij (t) O1:t) <ref:2608.04817#pg9> This state is constructed by continuously fusing four heterogeneous streams: delayed observations describe a past rather than a present state <ref:2608. p. 3

Where the paper says it

4 shows how the KCDT turns this environment into an Operational Knowledge State by fusing four heterogeneous streams. Delayed observations describe a past rather than a present state.

04817#pg10>, communication physics constrains what that state can plausibly be independent of any single observation <ref:2608. p. 3
Where the paper says it

Delayed observations describe a past rather than a present state. Communication physics constrains what that state can plausibly be independent of any single observation.

04817#pg10>, learned models capture the residual behavior that physics cannot explain together with their own predictive uncertainty <ref:2608. p. 3
Where the paper says it

Communication physics constrains what that state can plausibly be independent of any single observation. Learned models capture the residual behavior that physics cannot explain together with their own predictive uncertainty.

04817#pg10>, and mission objectives specify how precisely each part of the environment must be known <ref:2608. p. 3
Where the paper says it

Learned models capture the residual behavior that physics cannot explain together with their own predictive uncertainty. Mission objectives specify how precisely each part of the environment must be known.

04817#pg10>

Consistency Properties of the Framework

The proposed inference mechanism specializes established Gaussian fusion and covariance propagation results to the KCDT, stating that informative evidence should reduce posterior uncertainty, whereas uncertainty should not decrease during an observation gap without new evidence <ref:2608.04817#pg10> Furthermore, Knowledge Entropy is distinct from AoII because it quantifies posterior uncertainty when the true state is not promptly observable <ref:2608. p. 8

Where the paper says it

Knowledge Entropy is also distinct from AoII [32]. AoII penalizes the duration and severity of a mismatch between an estimate and the underlying source state, whereas entropy quantifies posterior uncertainty when the true state is not promptly observable.

04817#pg10> The framework also establishes consistency conditions for standard Gaussian information-form fusion and covariance prediction as the inference layer <ref:2608. p. 0
Where the paper says it

To quantify what is known, how uncertain it remains, and how current it is, we define two quantities over the fused Bayesian belief maintained for each communication link: Knowledge Entropy, the differential entropy of the posterior belief, and Knowledge Freshness, an Age of Information (AoI)-derived clock associated with the most recently assimilated evidence. Using standard Gaussian information-form fusion and covariance prediction as the inference layer, we establish the consistency conditions required by the proposed framework and demonstrate the operational significance of the resulting belief, entropy, and freshness through simulations and comparisons with flight data from the Longjiang-2 lunar microsatellite.

04817#pg10>

Knowledge-to-Decision Interface

The Operational Knowledge State becomes actionable through an admissibility-and-risk interface defined by the set of admissible links Eadm(t) where uncertainty and evidence age satisfy mission requirements <ref:2608.04817#pg11> For every admissible link, a normalized knowledge-aware cost cij(t) is defined as a function of propagation delay, entropy, freshness, and SNR <ref:2608. p. 6

Where the paper says it

A neural network estimates that residual as follows bϵij (t) = rθ (hij (t)), (10) where rθ(·) is the residual network with weights θ, and hij (t) is the observation history for link (i, j), defined as hij (t) = {zij (t1), zij (t2), . , zij (tn)} , (11) with zij (tk) the observation received at time tk ≤ t, comprising every delayed SNR, Doppler, and delay measurement collected so far.

04817#pg14> This leads to concrete reference policies for adaptive routing, proactive link switching, and uncertainty-aware resource allocation <ref:2608. p. 4
Where the paper says it

Long propagation delays Intermittent and obstructed propagation Far-side moon mission Resource constraint link Sat1 Sat2 LEO2 LEO1 Ground control Station Robotic rover Lunar orbit Lunar Sat1 Lunar Sat2 Earth orbit FIGURE 3: The cislunar ecosystem, comprising terrestrial ground stations, relay satellites, the lunar gateway, and robotic explorers, represented as a dynamic graph subject to intermittent visibility, resource constraints, and heterogeneous communication technologies. Operational Knowledge State Delayed Observations past sensing & telemetry, propagation latency Communication Physics propagation loss, Doppler shift, orbital geometry Uncertainty Quantification propagation & combination of uncertainty across all sources Mission Objectives requisite entropy & freshness per infrastructure element Learned Models residual behaviour, predictive uncertainty KnowledgeCentric Reasoning & Decision Making Adaptive Routing around degraded links Proactive Link Switching ahead of predicted loss Resource Allocation hedged against uncertainty Analytics Dashboard interpretable reasoning trace Autonomous Actions mission success: PMC ↑, NSP↑, VoK↑ decisions act back on the cislunar infrastructure, the resulting behavior becomes the next set of delayed observations FIGURE 4: The knowledge-centric digital twin continuously fuses delayed observations, communication physics, learned models, mission objectives, and uncertainty quantification into a single evolving operational knowledge state.

04817#pg14>

Validation Against Flight Data

The framework was evaluated using Monte Carlo simulations under the environment and protocol described in Section E, and a physically grounded comparison based on Longjiang-2 flight data <ref:2608.04817#pg18> The results show that distributed observations produced more informative Operational Knowledge than any individual communication source Furthermore, the knowledge-centric belief increased by 0. p. 1

Where the paper says it

: A Knowledge-Centric Communication For Autonomous Cislunar Networks: IEEE Open Journal of the Communications Society COMMUNICATION 1948 1990s – 2010s 2020s this work NEW Classical Communication Network-Centric Communication Semantic Communication Knowledge-Centric Communication faithful symbol reproduction metric: bit error rate reliable delivery under delay & disruption metric: throughput, AoI preservation of meaning & task relevance metric: semantic similarity reasoning over what is known, and not known metric: entropy, freshness DIGITAL TWIN ±σ Virtual Counterpart Predictive & Bidirectional AI-Native, Distributed Agentic / Cognitive Twin mirrors a physical asset across its lifecycle hybrid physics–data models, continual updating, UQ & V&V knowledge-graph-based architectures for cyber–physical systems twin contributes to the decision loop, not only feeds it ONBOARD AI Onboard Compression & Preprocessing Distributed Edge Intelligence Foundation Models, Space Cloud onboard processing limited to data compression collaborative sensing, federated learning, autonomous inference onboard foundation models embedded in the comms infrastructure communication frameworks optimize channels, routing, coding, and resource allocation • digital twins represent, synchronize, and predict physical systems • onboard intelligence extracts actionable information from sensed data a shared assumption, rarely stated autonomous decisions can be supported by sufficiently accurate operational information – i. , a subsystem can eventually reconcile its internal state against ground truth breaks down Earth – Moon infrastructure breaks this assumption • Propagation delay is bounded below by the physics of light travel time • Intermittent visibility is set by orbital geometry, not by deployment choice • Decisions must often execute before any corroborating observation can arrive manufacturing, aerospace-sustainment, and Earth-system twins can treat synchronization delay as an engineering nuisance, eventually reconciled against ground truth – in cislunar space that reconciliation is structurally unavailable, not a modelling simplification Autonomous communication decisions must instead integrate delayed observations, communication physics, predictive models, mission objectives, and quantified uncertainty into what the infrastructure currently knows, and does not know, about itself and its environment (this work) FIGURE 1: Three research threads converging on autonomy.

176 nats during the anomaly and remained elevated briefly afterward, which is consistent with reports that the jump corrupted some, but not all, of the affected packets <ref:2608.04817#pg19>

Conclusion and Future Directions

The framework suggests a shift from state-centric communication to knowledge-centric reasoning realized through the KCDT <ref:2608.04817#pg4> The framework is a starting point rather than a complete solution, and the shift from state-centric to knowledge-centric reasoning is expected to apply beyond Earth to Moon communications where autonomous systems must act on what they currently know rather than on what may eventually become known <ref:2608. p. 1

Where the paper says it

: A Knowledge-Centric Communication For Autonomous Cislunar Networks: IEEE Open Journal of the Communications Society COMMUNICATION 1948 1990s – 2010s 2020s this work NEW Classical Communication Network-Centric Communication Semantic Communication Knowledge-Centric Communication faithful symbol reproduction metric: bit error rate reliable delivery under delay & disruption metric: throughput, AoI preservation of meaning & task relevance metric: semantic similarity reasoning over what is known, and not known metric: entropy, freshness DIGITAL TWIN ±σ Virtual Counterpart Predictive & Bidirectional AI-Native, Distributed Agentic / Cognitive Twin mirrors a physical asset across its lifecycle hybrid physics–data models, continual updating, UQ & V&V knowledge-graph-based architectures for cyber–physical systems twin contributes to the decision loop, not only feeds it ONBOARD AI Onboard Compression & Preprocessing Distributed Edge Intelligence Foundation Models, Space Cloud onboard processing limited to data compression collaborative sensing, federated learning, autonomous inference onboard foundation models embedded in the comms infrastructure communication frameworks optimize channels, routing, coding, and resource allocation • digital twins represent, synchronize, and predict physical systems • onboard intelligence extracts actionable information from sensed data a shared assumption, rarely stated autonomous decisions can be supported by sufficiently accurate operational information – i. , a subsystem can eventually reconcile its internal state against ground truth breaks down Earth – Moon infrastructure breaks this assumption • Propagation delay is bounded below by the physics of light travel time • Intermittent visibility is set by orbital geometry, not by deployment choice • Decisions must often execute before any corroborating observation can arrive manufacturing, aerospace-sustainment, and Earth-system twins can treat synchronization delay as an engineering nuisance, eventually reconciled against ground truth – in cislunar space that reconciliation is structurally unavailable, not a modelling simplification Autonomous communication decisions must instead integrate delayed observations, communication physics, predictive models, mission objectives, and quantified uncertainty into what the infrastructure currently knows, and does not know, about itself and its environment (this work) FIGURE 1: Three research threads converging on autonomy.

04817#pg22> The paper opens a research direction in which the representation, quantification, and lifecycle of operational knowledge become as central to communication theory as the traditional concept of the bit <ref:2608. p. 1
Where the paper says it

: A Knowledge-Centric Communication For Autonomous Cislunar Networks: IEEE Open Journal of the Communications Society COMMUNICATION 1948 1990s – 2010s 2020s this work NEW Classical Communication Network-Centric Communication Semantic Communication Knowledge-Centric Communication faithful symbol reproduction metric: bit error rate reliable delivery under delay & disruption metric: throughput, AoI preservation of meaning & task relevance metric: semantic similarity reasoning over what is known, and not known metric: entropy, freshness DIGITAL TWIN ±σ Virtual Counterpart Predictive & Bidirectional AI-Native, Distributed Agentic / Cognitive Twin mirrors a physical asset across its lifecycle hybrid physics–data models, continual updating, UQ & V&V knowledge-graph-based architectures for cyber–physical systems twin contributes to the decision loop, not only feeds it ONBOARD AI Onboard Compression & Preprocessing Distributed Edge Intelligence Foundation Models, Space Cloud onboard processing limited to data compression collaborative sensing, federated learning, autonomous inference onboard foundation models embedded in the comms infrastructure communication frameworks optimize channels, routing, coding, and resource allocation • digital twins represent, synchronize, and predict physical systems • onboard intelligence extracts actionable information from sensed data a shared assumption, rarely stated autonomous decisions can be supported by sufficiently accurate operational information – i. , a subsystem can eventually reconcile its internal state against ground truth breaks down Earth – Moon infrastructure breaks this assumption • Propagation delay is bounded below by the physics of light travel time • Intermittent visibility is set by orbital geometry, not by deployment choice • Decisions must often execute before any corroborating observation can arrive manufacturing, aerospace-sustainment, and Earth-system twins can treat synchronization delay as an engineering nuisance, eventually reconciled against ground truth – in cislunar space that reconciliation is structurally unavailable, not a modelling simplification Autonomous communication decisions must instead integrate delayed observations, communication physics, predictive models, mission objectives, and quantified uncertainty into what the infrastructure currently knows, and does not know, about itself and its environment (this work) FIGURE 1: Three research threads converging on autonomy.

04817#pg22>

References

[1] National Aeronautics and Space Administration, “Moon to Mars architecture: Lunar communications, positioning, navigation, and timing (CPNT) architecture,” NASA, Washington, DC, USA, Tech. Rep., 2023

[2] S. G. C¸ etin, B. Donmez, and G. K. Kurt, “Advancing lunar communication through interdomain space networks and dynamic orchestration,” IEEE Transactions on Aerospace and Electronic Systems, vol. p. 1

Where the paper says it

: A Knowledge-Centric Communication For Autonomous Cislunar Networks: IEEE Open Journal of the Communications Society COMMUNICATION 1948 1990s – 2010s 2020s this work NEW Classical Communication Network-Centric Communication Semantic Communication Knowledge-Centric Communication faithful symbol reproduction metric: bit error rate reliable delivery under delay & disruption metric: throughput, AoI preservation of meaning & task relevance metric: semantic similarity reasoning over what is known, and not known metric: entropy, freshness DIGITAL TWIN ±σ Virtual Counterpart Predictive & Bidirectional AI-Native, Distributed Agentic / Cognitive Twin mirrors a physical asset across its lifecycle hybrid physics–data models, continual updating, UQ & V&V knowledge-graph-based architectures for cyber–physical systems twin contributes to the decision loop, not only feeds it ONBOARD AI Onboard Compression & Preprocessing Distributed Edge Intelligence Foundation Models, Space Cloud onboard processing limited to data compression collaborative sensing, federated learning, autonomous inference onboard foundation models embedded in the comms infrastructure communication frameworks optimize channels, routing, coding, and resource allocation • digital twins represent, synchronize, and predict physical systems • onboard intelligence extracts actionable information from sensed data a shared assumption, rarely stated autonomous decisions can be supported by sufficiently accurate operational information – i. , a subsystem can eventually reconcile its internal state against ground truth breaks down Earth – Moon infrastructure breaks this assumption • Propagation delay is bounded below by the physics of light travel time • Intermittent visibility is set by orbital geometry, not by deployment choice • Decisions must often execute before any corroborating observation can arrive manufacturing, aerospace-sustainment, and Earth-system twins can treat synchronization delay as an engineering nuisance, eventually reconciled against ground truth – in cislunar space that reconciliation is structurally unavailable, not a modelling simplification Autonomous communication decisions must instead integrate delayed observations, communication physics, predictive models, mission objectives, and quantified uncertainty into what the infrastructure currently knows, and does not know, about itself and its environment (this work) FIGURE 1: Three research threads converging on autonomy.

62, 2026

[3] A. Schlesinger, B. M. Willman, L. Pitts, S. R. Davidson, and W. A. Pohlchuck, “Delay/Disruption Tolerant Networking for the International Space Station (ISS),” in 2017 IEEE Aerospace Conference, Mar. p. 0

Where the paper says it

Early work on Delay/Disruption Tolerant Networking (DTN) established the networking principles needed to operate under long propagation delays and intermittent connectivity [3]. Optical communication later showed that laser links can substantially increase deepspace data rates over conventional radio-frequency (RF) systems, an approach validated in operation by NASA’s Deep Space Optical Communications (DSOC) project [4], [5].

2017

[4] A. E. Velasco, S. M. Meenehan, M. W. Wright, E. Alerstam, J. P. Allmaras, K Andrews, W C Buehlman V Garkanian C M G Jones and M Srinivasan “Operational Results From the Deep Space Optical Communications (DSOC) Project Ground Laser Transmitter,” IEEE Journal of Selected Topics in Quantum Electronics, vol. p. 0

Where the paper says it

Early work on Delay/Disruption Tolerant Networking (DTN) established the networking principles needed to operate under long propagation delays and intermittent connectivity [3]. Optical communication later showed that laser links can substantially increase deepspace data rates over conventional radio-frequency (RF) systems, an approach validated in operation by NASA’s Deep Space Optical Communications (DSOC) project [4], [5].

32, no. 1: Advances in Free Space Laser

[5] N J P Laboratory and D Team “Deep space optical communications: Operational results,” NASA Technical Report, 2026

[6] S. G. C¸ etin, B. Donmez, and G. K Kurt “Advancing Lunar Communication Through Interdomain Space Networks and Dynamic Orchestration,” IEEE Transactions on Aerospace and Electronic Systems, vol. p. 1

Where the paper says it

: A Knowledge-Centric Communication For Autonomous Cislunar Networks: IEEE Open Journal of the Communications Society COMMUNICATION 1948 1990s – 2010s 2020s this work NEW Classical Communication Network-Centric Communication Semantic Communication Knowledge-Centric Communication faithful symbol reproduction metric: bit error rate reliable delivery under delay & disruption metric: throughput, AoI preservation of meaning & task relevance metric: semantic similarity reasoning over what is known, and not known metric: entropy, freshness DIGITAL TWIN ±σ Virtual Counterpart Predictive & Bidirectional AI-Native, Distributed Agentic / Cognitive Twin mirrors a physical asset across its lifecycle hybrid physics–data models, continual updating, UQ & V&V knowledge-graph-based architectures for cyber–physical systems twin contributes to the decision loop, not only feeds it ONBOARD AI Onboard Compression & Preprocessing Distributed Edge Intelligence Foundation Models, Space Cloud onboard processing limited to data compression collaborative sensing, federated learning, autonomous inference onboard foundation models embedded in the comms infrastructure communication frameworks optimize channels, routing, coding, and resource allocation • digital twins represent, synchronize, and predict physical systems • onboard intelligence extracts actionable information from sensed data a shared assumption, rarely stated autonomous decisions can be supported by sufficiently accurate operational information – i. , a subsystem can eventually reconcile its internal state against ground truth breaks down Earth – Moon infrastructure breaks this assumption • Propagation delay is bounded below by the physics of light travel time • Intermittent visibility is set by orbital geometry, not by deployment choice • Decisions must often execute before any corroborating observation can arrive manufacturing, aerospace-sustainment, and Earth-system twins can treat synchronization delay as an engineering nuisance, eventually reconciled against ground truth – in cislunar space that reconciliation is structurally unavailable, not a modelling simplification Autonomous communication decisions must instead integrate delayed observations, communication physics, predictive models, mission objectives, and quantified uncertainty into what the infrastructure currently knows, and does not know, about itself and its environment (this work) FIGURE 1: Three research threads converging on autonomy.

62, pp. 1582–1594, 2026

[7] L Fernandez, J A Ruiz-De-Azua, A Calveras, and A Camps “Assessing LoRa for Satellite-to-Earth Communications Considering the Impact of Ionospheric Scintillation,” IEEE Access, vol. 8, pp. 165 570–165 582, 2020

[8] Y Wei, C Chen, L Yao, H Zhang, F Dong, F Li, F Wei G Jiang F Bao Y Zhou Z Li C Shen J Zhang N Chi and J Shi “A Semantic-Empowered Free-Space Optical Communication System With Turbulence-Resilient Vector Beams,” IEEE Transactions on Wireless Communications, vol. 25, pp. 11 530–11 545, 2026

[9] C E Shannon “A mathematical theory of communication,” Bell System Technical Journal, vol. 27, no. 3, pp. 379–423, 1948

[10] S Kaul R Yates and M Gruteser “Real-time status: How often should one update?” in IEEE INFOCOM, 2012

[11] D Gunduz Z Qin I E Aguerri H Dhillon Z Yang A Yener K K Wong and C B Chae “Beyond transmitting bits: Context, semantics, and task-oriented communications,” IEEE Journal on Selected Areas in Communications, vol. 41, no. 1, pp. 5–41, 2023

[12] Z Qin X Tao J Lu and G Y Li “Semantic communications: Principles and challenges,” arXiv preprint arXiv:2201.01389, 2021

[13] F Tao and Q Qi “Make more digital twins,” Nature, vol. 573, pp. 490–491, 2019

[14] K Willcox and B Segundo “The role of computational science in digital twins,” Nature Computational Science, vol. 4, pp. 166–168, 2024

[15] A Ferrari and K Willcox “Digital twins in mechanical and aerospace engineering,” Nature Computational Science, vol. 4, no. 3, pp. 178–183, 2024

[16] O San A Rasheed E Bozdemir and J Deng “The Evolution of Digital Twins from Reactive to Agentic Systems,” Nature Computational Science, vol. 6, no. 1, pp. 6–10, Jan. 2026 arXiv:2605. p. 0

Where the paper says it

org/licenses/by/4. 0/ VOLUME , 1 arXiv:2608.

26066 [cs.CE]

[17] L Zhou S Leng Y Liu Z Xiong and T Q S Quek “Digital Twins for Low-Altitude UAV Networks–Cooperation and Learning,” IEEE Transactions on Mobile Computing, vol. 25, no. 4, pp. 4839–4856, Apr. 2026

[18] A A Hamwi and M Mittal “FL-TWIN: a unified federated learning system for intrusion detection with digital twins modelling,” Scientific Reports, Jun. 2026

[19] X Li W Wang P Liu T Zhou and H Xia “Exploring Generative AI for Mitigating Digital Twin Deviation in Secure Vehicular Edge Computing,” IEEE Open Journal of the Communications Society, vol. 7, pp. 3046–3059, 2026

[20] M S Parwez S T Srivillibhutturu A Alshammari and H Olufowobi “Digital Twin–Guided AI Path Planning for ConnectivityAware Mobility,” IEEE Open Journal of the Communications Society, vol. 7, pp. 4259–4277, 2026

[21] L Wang X Ma J Fan F Li and Y Li “Digital Twin Architecture Design and Testbed Deployment for Satellite-Terrestrial Integrated Networks,” IEEE Open Journal of the Communications Society, vol. 7, pp. 153–168, 2026

[22] K Duran L Verda Cakir Y Yigit K Huseynov S Ram Kusu M Ali Erturk and B Canberk “Toward Digital Twin-as-a-Service (DTaaS) Platforms: A Survey on Architecture, Design Requirements, and Performance Metrics,” IEEE Communications Surveys & Tutorials, vol. 28, pp. 1845–1878, 2026

[23] D Barretta L Marini F Razzano G Guerrisi S Sandberg R Del Prete V Zancan E Mariotti G Furano L Manovi J Guo S Ullo N Longepe and G Meoni “Toward Real-Time Earth System Intelligence Through Advancements in Onboard AI Processing: A review,” IEEE Geoscience and Remote Sensing Magazine, vol. 14, no. 3, pp. 392–441, Jun. 2026

[24] X Lin L Kundu C Dick and S Velayutham “Embracing ai in 5g-advanced toward 6g: A joint 3gpp and o-ran perspective,” IEEE Communications Standards Magazine, vol. 7, no. 4, pp. 76–83, 2023

[25] K Guo P Qi S Mumtaz Y Huang A Nauman L Zhang and Q Wu “Reliable covert communication for integrated cognitive satellite–aerial–terrestrial networks with NOMA and poisson-distributed jammers,” IEEE Transactions on Communications, vol. 74, pp. 10 065–10 079, 2026

[26] M Wu K Guo Z Wang C Dong Y Liu Q Wu and Z Zheng “Federated learning-driven covert communication in satellite–terrestrial integrated networks: A privacy-preserving framework,” IEEE Journal on Selected Areas in Communications, vol. 44, pp. 2143–2157, 2026

[27] M Wu K Guo T A Tsiftsis S Mumtaz Z Wang Y Liu and Z Zheng “Hierarchical resource optimization for covert sagins: A stackelberg-matching game approach,” IEEE Transactions on Wireless Communications, vol. 25, pp. 20 558–20 573, 2026

[28] D A Vallado Fundamentals of Astrodynamics and Applications, 4th ed. Hawthorne, CA: Microcosm Press, 2013

[29] H Friis “A Note on a Simple Transmission Formula,” Proceedings of the IRE, vol. 34, no. 5, pp. 254–256, May 1946

[30] S Thrun W Burgard and D Fox Probabilistic Robotics, ser. Intelligent Robotics and Autonomous Agents. p. 0

Where the paper says it

NASA’s Communication, Positioning, Navigation and Timing (CPNT) architecture, LunaNet, and emerging interdomain lunar communication frameworks have since defined interoperable communication services for cislunar operations [1], [6]. Communication research has also expanded from propagation modeling and link-budget analysis toward semantic communications, intelligent routing, and AI-assisted systems that adapt to changing communication environments [2], [7], [8].

Cambridge, MA: MIT Press, 2005

[31] Barshalom Adaptive Estimation and Maneuvering Targets. John Wiley & Sons, Ltd ch. 11, pp. 421–490 [Online]

[32] A Maatouk S Kriouile M Assaad and A Ephremides “The age of incorrect information: A new performance metric for status updates,” IEEE/ACM Transactions on Networking, vol. 28, no. 5, pp. 2215–2228, 2020

[33] S Wang and S Hirche “Value of information in remote estimation subject to delay and packet dropouts,” in 62nd IEEE Conference on Decision and Control (CDC), 2023, pp. 3896–3902

[34] R A Horn and C R Johnson Matrix Analysis, 2nd ed. Cambridge; New York: Cambridge University Press, 2013

[35] R Bhatia Positive Definite Matrices, ser. Princeton Series in Applied Mathematics. p. 7

Where the paper says it

(26), whereas the assimilation time of its supporting evidence determines Knowledge Freshness through Eq. Mission-specific entropy and freshness requirements are subsequently applied through Eq.

Princeton, NJ: Princeton University Press, 2007

[36] S J Julier and J K Uhlmann “A non-divergent estimation algorithm in the presence of unknown correlations,” in Proc. American Control Conference (ACC), 1997, pp. p. 3

Where the paper says it

Section D formalizes representative routing, link-switching, and resource-allocation rules that consume the maintained knowledge state. These rules provide a concrete interface between estimation and control.

p. 3
Where the paper says it

Section D formalizes representative routing, link-switching, and resource-allocation rules that consume the maintained knowledge state. These rules provide a concrete interface between estimation and control.

2369–2373

[37] M Wei C Hu D Estevez M Tai Y Zhao J Huang C Bassa T Jan Dijkema X Cao and F Wang “Design and flight results of the VHF/UHF communication system of Longjiang lunar microsatellites,” Nature Communications, vol. 11, no. 1, p. 3425, Jul. 2020

[38] S J Julier and J K Uhlmann “A non-divergent estimation algorithm in the presence of unknown correlations,” in Proc. American Control Conference (ACC), 1997, pp.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Knowledge-Centric Digital Twin (KCDT) implementation allows autonomous decisions to integrate delayed observations, communication physics, predictive models, mission objectives, and quantified uncertainty into what the infrastructure currently knows and does not know about itself and its environment. This enables systems to make proactive link switching or uncertainty-aware resource allocation based on admissible-link definition rather than just instantaneous state.

  2. Knowledge Entropy quantifies how much remains unknown, allowing for mission-specific requirements where a safety-critical handover might require almost no entropy in the relevant link’s state. This shifts decision criteria from uniform precision to tailoring precision based on mission importance, as shown by the weightings used in calculating H(t).

  3. Knowledge Freshness measures how far evidence has decayed, defined as an Age of Information (AoI)-derived clock associated with the most recently assimilated evidence. This metric ensures that autonomous systems do not rely on stale data by providing a measure of how long ago the evidence underlying a belief was collected and, therefore, how far it may have decayed toward staleness.

  4. The KCDT framework allows for distributed knowledge fusion where Distributed observations produced more informative Operational Knowledge than any individual communication source, enabling mission success in scenarios where some regimes cannot achieve mission success by improving only one observation source.

  5. The system can generalize learned models across communication regimes, as shown by the residual model transferring from Regime A to Regime B without retraining when uncertainty is bounded, resolving extrapolation failure associated with the longest outages.

Abstract

Future lunar infrastructure requires communication as a persistent service rather than a mission specific capability. Communication, digital twin, and artificial intelligence (AI) research have advanced autonomy independently while largely assuming that system state can eventually be reconciled with ground truth. Cislunar communication violates this assumption because propagation delay is fundamentally limited by the speed of light, visibility is governed by orbital geometry, and autonomous decisions often precede confirming observations. Here we introduce a knowledge-centric communication framework in which a digital twin continuously integrates delayed observations, communication physics, learned models, uncertainty quantification, and mission objectives into an evolving Operational Knowledge State. To quantify what is known, how uncertain it remains, and how current it is, we define two quantities over the fused Bayesian belief maintained for each communication link: Knowledge Entropy, the differential entropy of the posterior belief, and Knowledge Freshness, which generalizes Age of Information (AoI) from a single information stream to the temporal validity of the fused Operational Knowledge State. We prove that distributed knowledge fusion systematically improves operational knowledge by reducing uncertainty and demonstrate its importance for mission success through simulations and comparisons with flight data from the Longjiang-2 lunar micro-satellite.

Sources

Related papers