Abstract
The Internet has evolved through successive architectural abstractions that enabled unprecedented scale, interoperability, and innovation. Packet-based networking enabled the reliable transport of bits; cloud-native systems enabled the orchestration of distributed computation. Today, the emergence of autonomous, learning-based systems introduces a new architectural challenge: intelligence is increasingly embedded directly into network control, computation, and decision-making, yet the Internet lacks a structural foundation for representing and exchanging meaning. In this paper, we argue that cognition alone: pattern recognition, prediction, and optimization, is insufficient for the next-generation of networked systems. As autonomous agents act across safety-critical and socio-technical domains, systems must not only compute and communicate, but also comprehend intent, context, and consequence. We introduce the concept of a Semantic Layer: a new architectural stratum that treats meaning as a first-class construct, enabling interpretive alignment, Semantic accountability, and intelligible autonomous behavior. We show that this evolution leads naturally to a Syntactic-Semantic Internet. The syntactic stack continues to transport bits, packets, and workloads with speed and reliability, while a parallel Semantic stack transports meaning, grounding, and consequence. We describe the structure of this Semantic stack—Semantic communication, a Semantic substrate, and an emerging Agentic Web, and draw explicit architectural parallels to TCP/IP and the World Wide Web. Finally, we examine current industry efforts, identify critical architectural gaps, and outline the engineering challenges required to make Semantic interoperability a global, interoperable infrastructure.
1 Introduction
Modern networks are becoming cognitive. Through advances in machine learning, distributed control, and edge–cloud convergence, systems can now sense their environments, detect patterns, optimize behavior, and act autonomously. These capabilities mark a fundamental shift from static, rule-driven infrastructures to adaptive, learning-based systems. Yet cognition, by itself, is insufficient for the systems we are now building ().
Cognitive systems operate primarily through statistical inference. They identify correlations, predict likely outcomes, and select actions that optimize learned objectives (). What they lack is the ability to determine why an action matters, whether it aligns with intent, or what consequences it may impose beyond local optimization. Two messages may be identical in structure and statistically similar in content, yet differ profoundly in meaning: one may represent a routine update, the other a safety-critical constraint. Without explicit Semantic grounding, a cognitive system has no architectural basis for distinguishing between them.
As autonomous systems scale and begin interacting with other autonomous systems, these limitations compound. Individually well-behaved agents may diverge over time as models adapt, assumptions shift, and local contexts evolve. Semantic drift accumulates (). Interpretations that were once aligned gradually fragment, leading to miscoordination, contradictory actions, and emergent failure modes. These failures are not computational bugs; they are architectural consequences of systems that lack shared structures for meaning.
A historical parallel clarifies the challenge. Before the advent of TCP/IP, early networks were sophisticated yet isolated. Each worked well internally, but interconnection failed because there was no shared architecture for addressing, routing, and transport. The problem was not intelligence or capability, but the absence of universally agreed-upon constructs. Cognitive systems today occupy a similar position. Without a structural layer for representing, aligning, and exchanging meaning, autonomous systems remain islands of intelligence rather than participants in coherent multi-agent ecosystems Cognition enables prediction but not interpretation, behavior but not understanding, action but not explanation (). Emerging autonomous infrastructures require more. They require comprehension: the ability to situate information within shared intent, contextual constraints, and anticipated consequences. Achieving this shift demands an architectural response, not merely better models (). That response is the Semantic Layer.
The Semantic layer treats meaning as a first-class engineering construct. By explicitly modeling intent, context, constraints, and anticipated consequences, it enables networks not only to transport data, but also to support shared understanding, interpretive alignment, and accountability. In doing so, the Semantic layer provides a structural foundation for autonomous collaboration across heterogeneous systems, application domains, and temporal scales.
Within the Semantic layer, Semantic communication constitutes the most fundamental capability. Unlike conventional communication paradigms that prioritize syntactic accuracy and bit-level fidelity, Semantic communication focuses on interpretive fidelity, i.e., whether the task-relevant Semantic content of information is preserved at the receiver (). Through task- and intent-aware representations, explicit contextual grounding, and structured handling of uncertainty, Semantic communication allows systems to maintain consistent understanding even under bandwidth constraints, model heterogeneity, and dynamically evolving environments. The introduction of Semantic communication therefore enables the Semantic layer to move from a conceptual abstraction toward an implementable architectural component, forming the basis for higher-level functions such as Semantic alignment, Semantic routing, and multi-agent coordination ().
2 Related work and positioning
Recent years have witnessed increasing interest in enabling Semantic-aware and autonomous networking, particularly in the context of 6G systems and large-scale distributed intelligence. This section reviews key developments across Semantic communication, intent-driven networking, and agent-based systems, and positions our work within this evolving landscape.
2.1 Semantic Communication and 6G initiatives
Semantic Communication has emerged as a promising paradigm that shifts the objective of communication from bit-level fidelity to Semantic fidelity. A growing body of work has explored task-oriented encoding, knowledge-aware transmission, and meaning-preserving representations. In parallel, large-scale research initiatives, including Horizon Europe programs () () (), the United States NSF RINGS () (), and VINES initiatives (), have investigated Semantic-aware networking as a foundational component of future 6G systems.
In addition, the International Telecommunication Union (ITU) has formalized the concept of Semantic-aware networking in its 2023 report on requirements for future networks (). These efforts collectively demonstrate a strong momentum toward embedding Semantic awareness into communication systems.
However, most existing works primarily focus on the physical and link-layer aspects of Semantic communication, such as encoding, compression, and goal-oriented transmission. The integration of Semantic representations into higher-layer networking functions, such as routing, coordination, and system-level interoperability, remains limited.
2.2 Intent-driven networking and autonomous systems
Intent-driven networking (IDN) has introduced a paradigm in which high-level user or system intents are translated into network configurations and operational policies. Recent advances in AI-driven network automation, closed-loop control, and autonomous network management (e.g., TM Forum Autonomous Networks) have demonstrated how intent abstraction can enable self-optimizing and self-adaptive systems ().
Moreover, many works have explored the role of Semantic representations in enabling intent interpretation and execution, particularly in the context of AI-driven data management and network orchestration. These approaches highlight the importance of bridging the gap between high-level intent and low-level system actions () ().
Despite these advances, intent-driven networking largely focuses on intra-domain automation and control. It lacks a general architectural framework for representing, exchanging, and aligning meaning across heterogeneous autonomous systems at Internet scale.
2.3 Agent-based systems and emerging agentic web
The emergence of large language models and autonomous agents has accelerated the development of agent-based communication frameworks. Recent industry and research efforts, such as Anthropic’s Model Context Protocol (MCP), Google’s Agent-to-Agent (A2A) protocols, IBM’s Agent Communication Protocol (ACP), and Cisco’s Internet of Agents, explore mechanisms for agent discovery, capability advertisement, and multi-agent coordination ().
These systems provide important building blocks for an emerging “Agentic Web,” where autonomous agents interact, negotiate, and collaborate to accomplish complex tasks. In parallel, research on knowledge graphs, ontologies, and the Semantic Web continues to provide foundational tools for representing structured meaning () ().
However, current agent-based frameworks remain largely ecosystem-specific and lack interoperability at a global scale. In particular, they do not yet provide a universal substrate for Semantic alignment, grounding negotiation, and provenance-aware interaction across heterogeneous agents.
2.4 Positioning of this work
Taken together, existing efforts in Semantic communication, intent-driven networking, and agent-based systems address important but largely isolated aspects of Semantic-aware networking. While they demonstrate the feasibility and importance of incorporating meaning into communication and system design, they do not converge into a unified architectural framework.
In this work, we aim to bridge this gap by proposing a Syntactic–Semantic Internet architecture, in which a parallel Semantic stack complements the traditional syntactic stack. Unlike prior approaches that focus on individual layers or components, our framework integrates Semantic communication, Semantic routing, grounding, and agent interaction into a coherent Internet-scale architecture.
To the best of our knowledge, this is among the first works to systematically unify these elements into a layered architectural model, enabling meaning to be represented, transmitted, and operationalized as a first-class construct across networked autonomous systems.
3 Semantic infrastructure for the Syntactic–Semantic internet
3.1 From data to meaningful action
To support comprehension at scale, networks must carry not only information but the significance of that information. Just as packets gave structure to the movement of bits, and workloads gave structure to the movement of computation, the Semantic Layer gives structure to the movement of meaning.
At its foundation, the Semantic Layer grounds information through contextual frames that capture the circumstances in which a message was created, the intent behind it, the constraints under which it should be interpreted, and the anticipated impact of acting upon it. These frames allow systems to distinguish superficially similar events that diverge fundamentally in meaning. They bind signals to concepts, concepts to intent, and intent to consequence.
The proposed Semantic Layer builds upon decades of research in knowledge representation and the Semantic Web. Technologies such as RDF, OWL ontologies, knowledge graphs, and Linked Data provide standardized mechanisms for representing structured Semantic relationships across distributed information systems. These frameworks enable machines to encode and query meaning at the data level (). However, most Semantic Web technologies operate primarily within data management and information integration contexts. The architecture proposed in this work extends these principles to the networking and systems level, enabling Semantic information to participate directly in communication, routing, coordination, and decision-making across autonomous systems.
The Semantic Layer also enables systems to negotiate meaning. When interpretive assumptions diverge or when information lacks sufficient grounding, systems must request clarification, renegotiate context, or decline to act. These behaviors are enabled not by cognitive heuristics but by structures that support interpretive alignment. The Semantic Layer therefore enforces the conditions under which comprehension can occur.
Finally, the Semantic Layer embeds the rationale behind actions. Cognitive models may output an action or a result without explanation. Semantic systems must expose why they acted: the intent, assumptions, grounding, provenance, and anticipated consequences behind each decision. This transparency transforms intelligent behavior into intelligible behavior ().
3.2 From Semantic concept to Semantic infrastructure
Concepts alone do not scale; infrastructure does. The Internet’s success is rooted in its architecture: abstractions, layers, and protocols that operationalized universal connectivity. The Semantic Layer requires the same discipline. To transform meaning from a localized property of individual models into a globally exchangeable resource, we must engineer a Semantic infrastructure composed of three tiers: Semantic communication, the Semantic substrate, and the Agentic Web in Figure 1.
FIGURE 1
Semantic communication: The Semantic communication layer redefines how information is represented and transmitted. Rather than treating data as anonymous symbols, it encodes each message within grounding frames that capture its intent, context, and expected consequences (). Semantic representations describe abstract meaning, autonomous systems must ultimately relate these abstractions to physical reality, grounding frames provide the interface between symbolic Semantics and operational context by mapping Semantic concepts to sensor measurements, environmental states, or actuation parameters. Representation becomes adaptive to the interpretive needs of the receiving agent. Messages emphasize critical Semantic features and elide irrelevant detail. This foundation mirrors the role of the physical and link layers in the syntactic stack, providing the channels through which meaningful information passes. Errors are defined not as bit-level distortions but as divergences in meaning. Semantic communication provides the mechanisms by which meaning begins its movement across the system ().
Semantic substrate: Above Semantic communication lies the Semantic substrate: the architectural equivalent of TCP/IP for meaning. The substrate establishes the universal structures necessary for meaning to move across heterogeneous systems. It defines identities for agents, the contexts within which interactions occur, and the envelopes that carry meaning across interpretive boundaries. In this substrate, meaning is not simply passed from one actor to another; it is negotiated, aligned, stabilized, and attested. Routing becomes an interpretive act (). Instead of forwarding messages based on topology, the Semantic substrate forwards meaning based on conceptual compatibility, capability, grounding alignment, and trust provenance. Systems become reachable not because they occupy a location in a network but because they possess the competence to interpret meaning correctly. Provenance is no longer metadata; it is a fundamental component of meaning. Drift detection becomes routine, enabling the substrate to identify when previously aligned interpretive structures begin to diverge ().
Agentic Web: Above the substrate lies the Agentic Web, which serves the same role for autonomous systems that the World Wide Web served for human users. It provides naming, discovery, interaction, and coordination across a landscape of autonomous agents. Just as the Web provided URLs, HTTP, HTML, and browsers that allowed humans to navigate information spaces, the Agentic Web provides Semantic registries, interaction protocols, shared grounding catalogs, and agentic runtimes that allow autonomous systems to navigate meaning spaces. It becomes the environment within which understanding is shared, negotiated, and acted upon. These three tiers are intentionally minimal abstractions, designed to permit heterogeneity above and below ().
3.3 Architectural parallels between the two stacks
As shown in
Figure 2, The syntactic stack is built around three central abstractions:
Physical/Link Layers: encode bits into signals.
Network/Transport Layers (TCP/IP): route packets across heterogeneous systems.
The Web: provides human-interpretable information spaces.
FIGURE 2
These abstractions energized the Internet’s scalability by separating concerns: bit transport, packet exchange, and information access.
The Semantic stack introduces corresponding abstractions:
Semantic Communication: encodes meaning relative to grounding frames.
The Semantic Substrate: routes meaning based on competency, context, and trust.
The Agentic Web: provides shared environments for agentic cooperation and Semantic interaction.
These Semantic abstractions preserve the same layering discipline, but operate in a different domain: meaning rather than bits. The stacks are therefore architectural counterparts, not alternatives.
3.4 How the two stacks operate together
3.4.1 The Syntactic-Semantic Internet functions through cooperative layering, not replacement
Syntactic channels provide connectivity, addressing, flow control, and packet integrity.
Semantic channels provide grounding synchronization, interpretive alignment, consequence management, and provenance guarantees.
A Semantic envelope is carried inside a packet but interpreted through the Semantic substrate. The packet ensures delivery; the Semantic envelope ensures understanding. Syntactic errors are corrected through retransmission; Semantic errors trigger renegotiation of meaning, realignment of grounding, or requests for clarification.
Semantic flow control differs from syntactic flow control: instead of regulating packet volumes, it regulates interpretive load, alignment stability, and the Semantic coherence of multi-agent exchanges.
DNS provides a naming system for resources; Semantic registries provide grounding authorities for meaning. These registries are not DNS, they do not map names to IP addresses, but they play an analogous role, mapping agents to competencies, grounding frames, and Semantic capabilities.
Over time, Semantic substrates, registries, and runtimes will become the invisible machinery of comprehension, just as routers and caches became the invisible machinery of connectivity. To make this architectural evolution explicit, we summarize the parallel roles in Table 1.
TABLE 1
| Dimension | Syntactic internet | Semantic internet |
|---|---|---|
| Foundational purpose | Transport bits | Transport meaning |
| Core abstraction | Packet | Semantic envelope |
| Representation focus | Symbols and data structures | Intent, context, grounding, consequence |
| Error model | Bit-level fidelity | Interpretive fidelity |
| Routing principle | Topology-driven packet forwarding | Capability-, context-, and intent-driven meaning routing |
| Identity | IP addresses | Agent identities with grounding and policy constraints |
| Interoperability basis | Shared syntax and standards | Shared interpretation, alignment, and semantic agreement |
| Trust model | Channel reliability, encryption, authentication | Provenance, rationale, semantic attestation, accountability |
| Application environment | Web (URLs, HTTP, HTML, browsers) | Agentic web (semantic registries, agent runtimes, A2A protocols) |
| Ultimate goal | Connectivity and data exchange | Coherence, comprehension, and accountable autonomy |
The syntactic-Semantic internet architecture.
3.4.2 Meaning in motion - Toward Semantic resilience
Just as bits face noise, loss, and corruption as they traverse networks, meanings face Semantic drift, misalignment, or loss of grounding as they traverse agents, domains, and time. Packets are largely stateless; Semantic envelopes which encapsulates data together with the contextual and interpretive information required for machine understanding are inherently stateful. They carry assumptions, intent, provenance, rationale, obligations, and policy bindings.
Preserving meaning therefore requires architectural mechanisms beyond reliable transport:
Semantic error detection
interpretive integrity checks
grounding preservation
Semantic retransmission
alignment protocols
consequence-sensitivity mechanisms
If the syntactic Internet engineered resilience for bits, the Semantic Internet must engineer resilience for understanding.
3.5 Industry initiatives and the architectural gaps that remain
Across industry, academia, and standards communities, momentum toward Semantic interoperability and agentic systems is unmistakable. Progress is occurring simultaneously in Semantic communication, agent runtimes, capability negotiation, and multi-agent coordination. Yet despite this activity, the current landscape mirrors the Internet before TCP/IP or the Web: innovation is advancing within individual layers, but without the architectural coherence required to form a global Semantic infrastructure.
At the foundation, research in Semantic communication is redefining how information should be encoded when the objective is comprehension rather than bitwise reconstruction. Some work in Part A of Section 2 illuminate what a Semantic counterpart to the physical and link layers might resemble, but they largely remain vertically isolated, without integration into Semantic routing, alignment, or agentic negotiation. In particular, in a syntactic network, routing decisions are based primarily on addresses and topology. In contrast, Semantic routing introduces additional decision dimensions based on meaning, intent, and capability compatibility.
Above this, multiple initiatives in Part C of Section 2 examine how autonomous systems discover one another, advertise capabilities, negotiate assumptions, and coordinate tasks. Each contributes essential primitives. None, however, yet constitutes the Semantic equivalent of TCP/IP: a minimal, universal substrate that allows meaning to move interoperably across heterogeneous agent ecosystems with shared guarantees of grounding, provenance, and interpretive fidelity.
At the application layer, early elements of an Agentic Web are beginning to emerge. Agent registries, discovery services, orchestration frameworks, and trust scaffolding are taking shape across open-source communities and industry consortia. These systems provide important local coherence, but they remain ecosystem-bound. What is still missing is a global Semantic registry, the true analogue of DNS, capable of resolving agent identities across domains and mapping them to competencies, grounding constraints, trust anchors, and policy contexts with Internet-scale universality.
Seen together, the pattern is clear. The components of a Semantic stack are advancing, but they are advancing in isolation. The industry is progressing within the Semantic stack, but not yet as a Semantic stack. What is absent is the architectural connective tissue: universal Semantic envelopes, cross-domain Semantic routing, and globally resolvable grounding and provenance.
This is precisely where the Syntactic-Semantic Internet becomes essential. Just as packet networking required a universal syntactic envelope and the Web required universal naming, Semantic interoperability requires a unifying architecture that binds communication, alignment, provenance, and interpretation into a coherent whole. Without this shared foundation, today’s agentic systems will remain powerful but fragmented. With it, they can converge into the next global substrate: an Internet capable not only of transporting data, but of transporting understanding.
3.6 Illustrative application scenarios
The Syntactic–Semantic Internet becomes particularly valuable in environments where heterogeneous autonomous systems must coordinate under dynamic and safety-critical conditions.
In autonomous transportation systems, Semantic communication could allow vehicles, infrastructure, and traffic management systems to exchange intent-aware messages. For example, rather than broadcasting raw sensor data alone, an autonomous vehicle may transmit a Semantic envelope describing an intended lane change together with contextual information about surrounding traffic conditions.
In industrial automation, Semantic interoperability could enable robots from different vendors to dynamically coordinate tasks within a manufacturing environment. Semantic routing mechanisms could direct task descriptions to machines with appropriate capabilities while grounding frames ensure that symbolic instructions map correctly to specific hardware configurations.
In smart infrastructure, distributed sensors, edge computing nodes, and control systems could exchange Semantically enriched information about environmental conditions, enabling coordinated responses across transportation, energy, and public safety systems.
4 Engineering challenges and the road ahead
Engineering the Semantic Layer requires solving challenges far more intricate than those faced during the evolution of the syntactic Internet. TCP/IP dealt with bit-level corruption, congestion collapse, heterogeneous link technologies, and global routing. The Semantic Internet inherits all these challenges and adds a new class of problems, those concerned not with information but with interpretation. Meaning, unlike bits, is inherently contextual, dynamic, and stateful. Preserving it across distributed agents therefore demands new forms of architectural discipline, new protocol categories, and new operational mechanisms.
A key aspect of the Semantic Layer lies in enabling capabilities such as Semantic routing, cross-agent alignment, and context-aware coordination. These aspects have attracted increasing attention in recent years, and a growing body of work has begun to explore related mechanisms, including semantic-aware communication strategies (), intent-driven networking (). However, despite these advances, existing efforts are typically developed in isolation, often tailored to specific application domains, system architectures, or operational assumptions. As a result, they lack a common way for representing and exchanging meaning across heterogeneous environments. In particular, current solutions do not yet provide a unified substrate for handling semantic routing decisions, maintaining interpretive alignment across agents with different grounding models, or supporting scalable coordination in open, Internet-scale settings.
Equally foundational is the development of alignment and grounding protocols, the mechanisms that allow agents to negotiate or re-establish shared understanding. Among them, Alignment protocols enable heterogeneous systems to reconcile Semantic interpretations through ontology mapping, shared references, or negotiation mechanisms. Over time, even well-aligned agents diverge, models drift, ontologies fragment, assumptions evolve, and interpretive frames adapt to local contexts. The Semantic Layer must make such drift detectable, measurable, and repairable. Alignment protocols must enable agents to query one another’s grounding, negotiate Semantic differences, resolve conflicts, and refuse actions when meaning cannot be safely reconstructed. In this sense, alignment behaves like Semantic flow control: not regulating packet rates but regulating the coherence and stability of interpretive exchange. Semantic alignment becomes the stabilizing mechanism that preserves coherence in multi-agent ecosystems, much as transport protocols stabilized packet flow across the early Internet.
Scalability remains a central concern, as Semantic representations may significantly increase metadata volume and reasoning complexity in large distributed systems. Efficient indexing, distributed knowledge management, and hierarchical Semantic routing strategies will be necessary to maintain performance at Internet scale.
Trust and provenance must also be elevated from syntactic protection to Semantic assurance. In today’s networks, trust is rooted in encryption, authentication, and channel integrity. In Semantic systems, trust must additionally account for the origin, rationale, grounding, and consequence of meaning. Each Semantic envelope must carry a verifiable lineage: who produced it, under what assumptions, with which model versions, subject to which obligations, and transformed by which interpretive steps. Without such provenance, autonomous systems cannot determine whether a meaning is valid, safe, compliant, or aligned with collective norms.
Operational management, the OAM of Semantic systems, requires new metrics and observability mechanisms. Traditional networks measure latency, throughput, jitter, and packet loss. Semantic networks must measure interpretive fidelity: the degree to which received meaning reflects sender intent. This requires new observability mechanisms capable of tracking Semantic drift, measuring grounding discrepancies, quantifying ambiguity, and evaluating interpretive uncertainty. Semantic retransmission may be needed not because packets were lost, but because understanding was.
Above all, standards must evolve to support these new constructs. Just as the syntactic Internet required standardized packet formats, addressing schemes, transport Semantics, and application-layer protocols, the Semantic Internet requires Semantic envelope specifications, grounding registries, agent-to-agent protocols, provenance formats, capability directories, and Semantic routing rules. These standards will emerge only through open governance, cross-disciplinary consensus, and iterative experimentation, echoing the design culture that shaped TCP/IP.
Interoperability testing will also take on a fundamentally new role. Instead of simply verifying whether two systems can exchange packets, future interoperability events must determine whether they can exchange meaning: whether agents interpret concepts consistently, negotiate grounding effectively, detect and repair drift, propagate provenance securely, and act according to shared norms. These tests will define the practical boundaries of Semantic interoperability, just as early TCP/IP bake-offs defined the boundaries of syntactic interoperability. In addition, computational constraints must be considered, particularly for edge devices and embedded autonomous systems. Lightweight Semantic representations and distributed reasoning architectures may be required to balance Semantic richness with real-time operational requirements.
Together, these challenges signal not merely the extension of existing architectures but the emergence of a fundamentally new one: a layer that binds meaning to action, aligns autonomous systems across domains, and enables comprehension to move across networks with the same reliability that packets move today. The engineering of this layer will shape not only how intelligent systems communicate, but also how they reason, cooperate, and remain accountable in a world where understanding itself moves across networks.
5 Conclusion
Across its history, the Internet has advanced by introducing architectural layers that abstract complexity while preserving interoperability. Packets stabilized global communication. Cloud-native abstractions stabilized distributed computation. Today, autonomous agents are stabilizing distributed intelligence, but intelligence alone is not sufficient for systems that increasingly act on the world.
As autonomy expands into safety-critical, economic, and socio-technical domains, systems must do more than compute and communicate. They must comprehend. Comprehension requires the ability to represent intent, situate information within context, reason about consequence, and remain accountable as meaning propagates across agents, domains, and time.
This paper has argued that these requirements cannot be met by extending the existing Internet stack alone. Instead, they necessitate a new architectural stratum: the Semantic Layer, which treats meaning itself as a first-class engineering construct. From this follows a natural and inevitable outcome: the Syntactic-Semantic Internet. The syntactic stack continues to ensure performance, reachability, and packet-level resilience. In parallel, the Semantic stack stabilizes the movement of meaning through grounding, provenance, interpretive alignment, and consequence awareness.
The two stacks are not competing paradigms but complementary infrastructures. One moves bits with fidelity; the other moves understanding with accountability. Together, they form the foundation for autonomous systems that can be trusted, reasoned about, and governed at planetary scale.
But meaning in motion introduces new vulnerabilities. Just as packets face loss and corruption, meaning faces Semantic drift, misalignment, and loss of grounding as it traverses agents and evolves over time. Semantic correctness cannot be defined by equivalence alone; it must be preserved as a stateful object carrying intent, assumptions, provenance, and obligation. Ensuring this demands new architectural mechanisms for Semantic error detection, alignment repair, grounding recovery, and consequence-aware renegotiation.
If the syntactic Internet engineered resilience for bits, the Semantic Internet must engineer resilience for understanding.
The Syntactic-Semantic Internet represents the next phase in the Internet’s evolution: from connectivity, to computation, to comprehension. The design and standardization of this Semantic infrastructure will shape how autonomous systems cooperate, how responsibility is assigned, and how meaning itself can be safely exchanged in an increasingly agentic world.
Statements
Author contributions
MT: Project administration, Conceptualization, Investigation, Writing – original draft, Funding acquisition. XL: Writing – review and editing, Methodology, Formal Analysis, Data curation, Resources. YS: Methodology, Supervision, Data curation, Resources, Writing – review and editing. MI: Writing – review and editing, Supervision, Resources, Visualization, Investigation, Funding acquisition, Validation.
Funding
The author(s) declared that financial support was not received for this work and/or its publication.
Conflict of interest
The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declared that generative AI was not used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
References
1
AbuelsaadT.AkkilD.DeyP.JagmohanA.VempatyA.KokkuR. (2024). Agent-e: From Autonomous Web Navigation to Foundational Design Principles in Agentic Systems. arXiv preprint arXiv:2407.13032.
2
ChengR.SunY.NiyatoD.ZhangL.ZhangL.ImranM. A. (2024). A wireless ai-generated content (aigc) provisioning framework empowered by semantic communication. IEEE Trans. Mob. Comput.24 (3), 2137–2150. 10.1109/tmc.2024.3493375
3
DorriA.KanhereS. S.JurdakR. (2018). Multi-agent systems: a survey. Ieee Access6, 28573–28593. 10.1109/access.2018.2831228
4
FonsecaJ.AdhaneG.TogiasK.OjedaJ.RoyS.FamitafreshiG.et al (2024). “Realizing intent-driven network management with tm forum standards,” in 2024 IEEE 29th International Workshop on Computer Aided Modeling and Design of Communication Links and Networks (CAMAD), 1–5.
5
HuangK.HughesC. (2025). “Agentic ai communication protocols and security,” in Securing AI Agents: Foundations, Frameworks, and Real-World Deployment (Springer), 81–110.
6
International Telecommunication Union (2023). Requirements of semantic-aware networking for future networks. ITU-T, Technical Report TR.Reqts-SANAvailable online at: https://www.itu.int/dms_pub/itu-t/opb/tut/T-TUT-SEMANTIC-2023-1-PDF-E.pdf (Accessed 11, 2023).
7
KouvakisV.TrevlakisS. E.BoulogeorgosA.-A. A. (2024). Semantic communications for image-based sign language transmission. IEEE Open J. Commun. Soc.5, 1088–1100. 10.1109/ojcoms.2024.3360191
8
LiY.SongL.ChenY.LiZ.ZhangX.WangX.et al (2020). “Learning dynamic routing for semantic segmentation,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 8553–8562.
9
LiY.ZhouF.YuanL.WuQ.Al-DhahirN. (2025). “Cognitive semantic communication: a new communication paradigm for 6G,” in IEEE Communications Magazine63 (6), 122–129. 10.1109/MCOM.005.2400366
10
LiuX.SunY.ChengR.XiaL.AbumarshoudH.ZhangL.et al (2025a). Knowledge-assisted privacy preserving in semantic communication. IEEE Wirel. Commun.32 (2), 76–83. 10.1109/mwc.001.2400315
11
LiuX.LiuY.TangH.ZhaoF.XiaL.SunY. (2025b). “Joint knowledge and power management for secure semantic communication networks,” in IEEE Transactions on Vehicular Technology75 (6), 10171–10184. 10.1109/TVT.2025.3636880
12
LuZ.LiR.LuK.ChenX.HossainE.ZhaoZ.et al (2023). Semantics-empowered communications: a tutorial-cum-survey. IEEE Commun. Surv. and Tutorials26 (1), 41–79. 10.1109/comst.2023.3333342
13
National Science Foundation (2025). Verticals-enabling intelligent network systems (vines). Available online at: https://www.nsf.gov/funding/opportunities/vines-verticals-enabling-intelligent-network-systems/506396/nsf25-539/solicitation.
14
PangL.YangC.ChenD.SongY.GuizaniM. (2020). A survey on intent-driven networks. IEEE Access8, 22862–22873. 10.1109/access.2020.2969208
15
ShahrakiA.TaherkordiA.HaugenØ.EliassenF. (2020). A survey and future directions on clustering: from wsns to iot and modern networking paradigms. IEEE Trans. Netw. Serv. Manag.18 (2), 2242–2274. 10.1109/tnsm.2020.3035315
16
SifakisJ.HarelD. (2023). Trustworthy autonomous system development. ACM Trans. Embed. Comput. Syst.22 (3), 1–24. 10.1145/3545178
17
StrinatiE. C.Di LorenzoP.SciancaleporeV.AijazA.KountourisM.GündüzD.et al (2024). “Goal-oriented and semantic communication in 6g ai-native networks: the 6g-goals approach,” in 2024 Joint European Conference on Networks and Communications 6G Summit (EuCNC/6G Summit), 1–6.
18
TangY.ZhaoC.WangJ.ZhangC.SunQ.ZhengW. X.et al (2022). Perception and navigation in autonomous systems in the era of learning: a survey. IEEE Trans. Neural Netw. Learn. Syst.34 (12), 9604–9624. 10.1109/TNNLS.2022.3167688
19
TrevlakisS. E.PappasN.BoulogeorgosA.-A. A. (2024). Toward natively intelligent semantic communications and networking. IEEE Open J. Commun. Soc.5, 1486–1503. 10.1109/ojcoms.2024.3371871
20
WangY.ChenM.SaadW.LuoT.CuiS.PoorH. V. (2021). “Performance optimization for semantic communications: an attention-based learning approach,” in 2021 IEEE Global Communications Conference (GLOBECOM), 1–6.
21
XiaoY.LiaoY.LiY.ShiG.PoorH. V.SaadW.et al (2024). Reasoning over the air: a reasoning-based implicit semantic-aware communication framework. IEEE Trans. Wirel. Commun.23 (4), 3839–3855. 10.1109/twc.2023.3312115
22
YangW.DuH.LiewZ. Q.LimW. Y. B.XiongZ.NiyatoD.et al (2022). Semantic communications for future internet: fundamentals, applications, and challenges. IEEE Commun. Surv. and Tutorials25 (1), 213–250. 10.1109/comst.2022.3223224
23
ZhaoF.SunY.LanJ.ZhangL.LiuX.ImranM. A. (2025). Adaptive semantic communication for uav/ugv cooperative path planning. arXiv Preprint arXiv:2510.06901. 10.1109/TVT.2026.3653997
24
ZhongF.WuY.YuH.WangG.LuZ. (2025). A benchmark dataset and semantics-guided detection network for spatial–temporal human actions in urban driving scenes. Pattern Recognit.158, 111035. 10.1016/j.patcog.2024.111035
Summary
Keywords
autonomous systems, Semantic architecture, Semantic communication, Semantic internet, Syntactic-Semantic internet
Citation
Tatipamula M, Liu X, Sun Y and Imran MA (2026) The Syntactic–Semantic internet: engineering infrastructures for autonomous systems. Front. Commun. Netw. 7:1812410. doi: 10.3389/frcmn.2026.1812410
Received
16 February 2026
Revised
05 June 2026
Accepted
05 June 2026
Published
01 July 2026
Volume
7 - 2026
Edited by
Mauro Femminella, University of Perugia, Italy
Reviewed by
Ben Hnaien Ilhem, The National Engineering School of Sousse (ENISo), Tunisia
Joao Pedro Fonseca, Onomondo Aps, Denmark
Updates
Copyright
© 2026 Tatipamula, Liu, Sun and Imran.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Muhammad Ali Imran, muhammad.imran@glasgow.ac.uk
Disclaimer
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.