Dirk Kutscher
Hong Kong University of Science and Technology (Guangzhou)
IEEE ICNP 2026 Full-Day Workshop
AgentNet 2026
Networking Foundations for Autonomous Agents
AgentNet explores whether large-scale agent ecosystems need a new networking thin waist for naming, discovery, trust, communication, delegation, provenance, and coordination across administrative domains.
This site starts with the accepted workshop proposal content and the confirmed logistics from the organizer email exchange. Additional details will be updated as the ICNP 2026 workshop schedule is finalized.
Organizers
Workshop Organizing Team
The initial workshop website follows the requested starting point: title, organizer list, and the material from the accepted call for papers onward.
Lan Wang
University of Memphis
Beichuan Zhang
University of Arizona
Lixia Zhang
University of California, Los Angeles
Call for Papers
Networking Foundations for Autonomous Agents
The emergence of AI agents introduces new requirements for networked systems. Agents are increasingly expected to discover services and other agents, delegate tasks, coordinate actions, exchange state, reason over provenance, and operate across administrative and trust boundaries. These interaction patterns go beyond traditional host-centric request-response communication and raise fundamental questions for network protocol design.
Current Internet architectures provide strong foundations for packet delivery, naming, transport, and secure communication channels, but they do not directly provide abstractions for agent identity, capability-based discovery, intent-driven communication, trust delegation, provenance-aware exchange, or accountable multi-agent coordination. As a result, many emerging agent systems rely on application-specific integration frameworks, centralized platforms, or ad hoc protocol conventions.
AgentNet 2026 invites original research, position papers, and early-stage work on the networking foundations of autonomous agent ecosystems. A central question for the workshop is whether large-scale agent ecosystems require a new architectural thin waist: a minimal interoperability substrate that allows heterogeneous agents, tools, models, platforms, and administrative domains to communicate and coordinate without requiring global agreement on application frameworks, reasoning engines, or trust authorities.
The workshop seeks to bring together researchers and practitioners from the network protocols, distributed systems, security, AI systems, information-centric networking, Web architecture, and federated systems to identify the key protocol abstractions, architectural principles, and deployment models for agent networking.
Topics of Interest
Research Directions
Topics of interest include, but are not limited to, the following areas.
- Thin-waist architectures for agent networking
- Minimal interoperability substrates for agent ecosystems
- Agent naming, identity, addressing, and persistence
- Capability-based discovery and routing
- Intent-driven, goal-oriented, and receiver-driven communication
- Transport protocols for agent-to-agent and agent-to-service interaction
- Multi-party, streaming, pub/sub, and asynchronous agent communication
- Trust delegation, authorization, and accountable action execution
- Provenance, authenticity, and verifiability of agent outputs
- Secure communication across independently governed domains
- Policy-aware and federated agent systems
- Multi-agent coordination, shared state, and synchronization
- Consistency models for distributed agent state
- Naming and discovery across heterogeneous agent platforms
- Relationships to ICN, DNS, Web architecture, and service discovery
- Network support for dynamic agent instantiation and migration
- Edge, cloud, and local-inference architectures for agent ecosystems
- Measurement, benchmarking, and performance evaluation of agent systems
- Scalability, robustness, and failure handling in large-scale agent ecosystems
- Real-world deployments, case studies, privacy, governance, and regulation
Submission Types
What to Submit
Full Research Papers
Original research results, systems, protocols, measurements, or evaluations.
Short Papers
Early-stage ideas, preliminary results, or focused technical contributions.
Position Papers
Architectural viewpoints, research agendas, open problems, or critical perspectives.
Experience Papers
Practical systems, implementation experience, operational lessons, or case studies.
Paper Format and Review
Requirements
Formatting
Submitted papers must be original and unpublished work and should follow the standard ICNP formatting guidelines.
Full papers may be up to 6 pages, excluding references. Short, position, and experience papers may be up to 4 pages, excluding references.
Review Process
All submissions will be reviewed by the technical program committee. Reviews will consider technical quality, originality, relevance to the workshop scope, clarity of presentation, and potential to stimulate discussion.
At least one author of each accepted paper must register for and present the work at the workshop.
Important Dates
Schedule Snapshot
Confirmed dates are shown directly. Remaining deadlines will be updated once they are finalized with ICNP 2026.
The organizers are still coordinating the paper due date.
The acceptance notification schedule will be posted here.
ICNP 2026 workshop organizers were asked to meet the conference camera-ready schedule.
Tempe, Arizona, USA.