Intent Based Networking and Automated Configuration Synthesis

39 concepts31 papers1 proposal

Traditional network management is fundamentally procedural. To implement operational goals—such as ensuring low latency for video traffic, isolating secure financial data, or routing telemetry across satellite constellations—operators must manually script thousands of low-level device configurations, flow rules, and routing tables. As networks expand into heterogeneous cloud-edge continuums, this manual approach creates an unsustainable operational burden. As noted by Adam Barker et al. in Cloud Services Brokerage: A Survey and Research Roadmap, infrastructure management has become a "paradox of choice," where users struggle to map high-level business goals to complex, heterogeneous hardware specifications.

Intent-Based Networking (IBN) solves this problem by shifting management from a procedural model to a declarative one. Instead of defining how a network should route packets, operators declare what the network should achieve. The system then automatically synthesizes, deploys, and maintains the required configurations.

Abstracting Requirements through Semantic Intents

How can an application or operator communicate its goals without knowing the underlying hardware topology? IBN introduces a semantic abstraction layer that decouples high-level requirements from low-level execution.

Early frameworks achieved this by using structured domain languages. For example, Yehia Elkhatib et al. proposed a linguistic structure of verbs and modifiers in Charting an Intent Driven Network, allowing applications to signal traffic semantics without exposing encrypted payloads. By specifying abstract intents rather than hard-coded endpoints, the underlying infrastructure can dynamically resolve and bind service providers based on real-time network conditions, a mechanism demonstrated by Abdessalam Elhabbash et al. in Adaptive Service Deployment using In-Network Mediation.

Recently, the field has shifted toward natural language processing (NLP) and Large Language Models (LLMs) to lower the barrier to entry. Xiaowen Chu et al. highlighted how combining NLP with deterministic finite automata can translate human-readable requirements into machine-executable policies in Guest Editorial: Interplay between Machine Learning and Networking Systems.

In highly dynamic environments, such as 6G and Integrated Satellite-Aerial-Terrestrial Networks (ISATNs), the authors of 6G智能业务网络:愿景、架构与关键技术 and Leveraging Large Language Models for Integrated Satellite-Aerial-Terrestrial Networks: Recent Advances and Future Directions argue that LLMs can act as intelligent orchestrators. These models parse abstract operator commands into technical scripts, automated beamforming configurations, or node positions. While Geng Sun et al. note in Large Language Model (LLM)-enabled Graphs in Dynamic Networking that reliably mapping abstract human objectives to precise low-level settings remains an active research challenge, the linguistic layer fundamentally changes the operator's role from a coder of configurations to an author of goals.

Automated Translation and Configuration Synthesis

An intent is useless without a mechanism to enforce it. The core engine of an IBN system is the intent compiler, which decomposes high-level Service Level Objectives (SLOs) into concrete, device-specific execution rules.

This translation operates by converting abstract constraints into formal mathematical or logical parameters that optimization solvers can process. For instance, updating network flows without causing loops or congestion is a high-stakes balancing act. Long Luo et al. demonstrate in Simplifying Flow Updates in Software-Defined Networks Using Atoman that high-level intent-based APIs can define desired network states (such as loop-freedom or bandwidth limits) as formal mathematical constraints. Automated solvers then handle the underlying optimization, bypassing manual heuristic development and drastically reducing scripting errors.

This automated synthesis applies across multiple networking domains:

Closed-Loop Verification and the Knowledge Plane

An intent is not a "fire-and-forget" command. Because network conditions change dynamically due to hardware failures, congestion, or security threats, an IBN system must operate as a closed loop. It must continuously observe the physical state of the network, verify it against the logical intent, and execute corrective actions if a deviation occurs.

This continuous verification requires a cognitive layer, often referred to as the "Knowledge Plane." As proposed by Pere Barlet-Ros et al. in Knowledge-Defined Networking, this mechanism uses a cognitive layer to ingest telemetry, build behavioral models, and issue high-level intents that a controller translates into hardware rules. This creates an automated feedback loop between observation and execution. Similarly, Dapeng Wu 0001 et al. describe a high-level intent parser that decomposes abstract objectives into discrete sub-tasks mapped to distributed computational engines, shifting the network's role from simple data transport to autonomous goal fulfillment in Vision and Challenges for Knowledge Centric Networking.

+-----------------------------------------------------------+
|                      Operator Intent                      |
|          "Maintain low latency & secure traffic"          |
+-----------------------------------------------------------+
                              |
                              v
+-----------------------------------------------------------+
|                 Intent Compiler / Parser                  |
|       Translates abstract goals to formal policies        |
+-----------------------------------------------------------+
                              |
                              v
+-----------------------------------------------------------+
|                    Knowledge Plane                        |
|   Synthesizes configurations & models network behavior    |
+-----------------------------------------------------------+
           |                                      ^
           v                                      |
+---------------------+                 +-------------------+
|  Physical Network   | --Telemetry-->  |   Digital Twin /  |
|  & Flow Deployment  |                 |  Monitoring State |
+---------------------+                 +-------------------+
           ^                                      |
           |                                      v
           +---------- Corrective Actions --------+
                    (Closed-Loop Automation)

To perform these "what-if" simulations safely without risking live infrastructure, modern frameworks rely on Network Digital Twins (NDTs). Yong Cui et al. highlight in Network Digital Twin how modeling existing configurations using graph neural networks enables automated template generation based on operator intent. In 6G radio access networks, Guangyi Liu 0001 et al. combine intent-driven management with digital twins to parse abstract goals and automatically generate functional parameters, moving closer to a "zero-touch" network capable of self-optimization and self-evolution 6G autonomous radio access network empowered by artificial intelligence and network digital twin.

Ultimately, this closed-loop orchestration is expanding across the entire cloud-edge continuum. Rustem Dautov et al. outline a roadmap in INTEND: Human-Like Intelligence for Intent-Based Data Operations in the Cognitive Computing Continuum where distributed systems prioritize high-level business objectives over manual technical scripting, allowing the environment to autonomously manage data placement and bandwidth adjustments across heterogeneous hardware.

To achieve full runtime autonomy, the monitoring infrastructure itself must be automated. As Niloy Saha et al. observe in MonArch: Network Slice Monitoring Architecture for Cloud Native 5G Deployments, automating the mapping between high-level monitoring requests and low-level container orchestration manifests is the final link required to eliminate manual intervention and close the loop entirely.

Key Foundational Work

Cloud Services Brokerage: A Survey and Research Roadmap (2015) — contributed 2 concept elements to this direction.
Merlin: A Language for Provisioning Network Resources (2014) — contributed 2 concept elements to this direction.
WebGC: browser-based gossiping (2014) — contributed 1 concept elements to this direction.
History & citationVersion 1 · 2026-10-04

Current · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_intent_based_networking_8a7e7de7,
  title = {intent-based networking},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/intent-based-networking/rdr_8a7e7de7e36f49b88a57703011ef9f7e},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-19},
}

This exact version (version 1, frozen text):

@misc{lacuna_intent_based_networking_8a7e7de7_v1,
  title = {intent-based networking},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/intent-based-networking/rdr_8a7e7de7e36f49b88a57703011ef9f7e/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-19},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake