Source-linked AI summary
Publish/subscribe-enabled software defined networking for efficient and scalable IoT communications
Akram Hakiri, Pascal Berthou, Aniruddha Gokhale, Slim Abdellatif
TL;DR
IoT communications must handle heterogeneous, mobile, resource-constrained devices and unresolved networking, QoS, standardization, and security challenges. The paper proposes combining SDN with DDS middleware, using DDS interfaces and SDN control to support network-agnostic IoT orchestration. The authors conclude that this symbiotic architecture enables agile and flexible network orchestration and can improve IoT communication.
Problem
IoT communication must accommodate heterogeneous, dynamic, resource-constrained devices while existing DDS capabilities are rarely propagated to the network layer and DDS is not itself a network-level solution.
Method
The paper combines SDN and DDS by adding a DDS northbound interface to an SDN controller and using DDS publish/subscribe middleware across smart objects and gateways.
Results
The proposed SDN-DDS architecture enables agile and flexible network orchestration for IoT systems.
Takeaways & Limitations
The architecture provides a network-agnostic approach intended to improve IoT communication and support standardized, modular, and adaptable deployments.
Abstract
from arXiv · showhide
- The Internet of Things (IoT) is the result of many different enabling technologies such as embedded systems, wireless sensor networks, cloud computing, big-data, etc. used to gather, process, infer, and transmit data. Integrating all these technologies requires a comprehensive and holistic research effort to address all the challenges imposed by these technologies, especially for sensing and delivering information from physical world to cloud-hosted services. In this paper, we outline the most important issues related to standardization efforts, mobility of objects, networking and gateway access, and QoS support. In particular, we describe a novel IoT network architecture that integrates Software Defined Networking (SDN) and the Object Management Group's Data Distribution Service (DDS) middleware. The proposed architecture will improve service delivery of IoT system and will bring flexibility to the network.
I. INTRODUCTION
IoT networks must support heterogeneous, dynamic, resource-constrained devices while remaining scalable, interoperable, predictable, and reprogrammable. The paper addresses limitations in existing middleware and networks by combining SDN with DDS.
- Requirements: IoT deployment involves heterogeneous smart objects whose connectivity, locations, speeds, and operational states can change dynamically.Devices may sleep, wake, connect, disconnect, or move while operating with limited computation, storage, and power.
- Networking gap: Existing IoT networking faces IPv4 exhaustion, heterogeneity, device-discovery, context-awareness, and resource-utilization challenges.The paper argues that current protocols and provisioning approaches lack abstractions and flexibility for changing IoT application needs.
- Related approaches: Web-of-Things approaches reduce heterogeneity through open web technologies but do not provide distributed, peer-to-peer publish/subscribe semantics.Security and scalability challenges also remain unresolved.
- Related approaches: Existing middleware contributions address upper-layer access and services without sufficiently addressing the underlying network, while DDS is not itself a network-level solution.DDS capabilities are rarely propagated to the network layer, limiting use across wide-area environments.
- Contribution: The paper combines SDN and DDS to create a protocol- and technology-independent abstraction addressing both network and distributed-systems concerns.The proposed combination is intended to support a more flexible IoT network architecture.
II. BACKGROUND
The background introduces SDN as a programmable, controller-based networking architecture and DDS as standardized middleware for interoperable publish/subscribe communication. It also situates both within ongoing networking-function virtualization and DDS standardization efforts.
- SDN and NFV: SDN separates network control from switches and uses a centralized controller with northbound application interfaces and southbound device protocols.OpenFlow uses software-defined flow tables to process and forward packets.
- SDN and NFV: NFV complements SDN by virtualizing network functions as software modules running on standardized IT infrastructure.These modules can be assembled or chained to create services.
- DDS: DDS is standardized middleware for interoperability among connected machines, enterprise systems, and mobile devices, spanning low-footprint devices to cloud platforms.It supports efficient bandwidth usage and agile orchestration through a flexible, modular structure.
- DDS: DDS models communication through domains, topics, publishers, data writers, subscribers, and data readers.A domain provides a virtual global data-space, while publishers and subscribers manage writers and readers.
- DDS evolution: DDS standardization has expanded from platform- and language-independent publish/subscribe mechanisms toward discovery through RTPS and planned SDN northbound integration.The standard also discusses dynamic topic types, remote method invocation, web integration, and security extensions.
III. NETWORK COMMUNICATION CHALLENGES IN IOT SYSTEMS
IoT architecture spans device, network, and application domains connected through heterogeneous access technologies and gateways. Gateways must integrate connectivity while managing resources, QoS, security, and data exchange.
- Architecture: IoT systems are organized into device, network, and application domains.The device domain contains smart devices and access networks, the network domain provides heterogeneous connectivity, and the application domain contains applications and cloud infrastructure.
- Device domain: Devices connect to the network directly through access networks such as Bluetooth and ZigBee or indirectly through gateways acting as network proxies.Gateways extend connectivity between device-side technologies and the broader network.
- Gateway requirements: IoT gateways must manage heterogeneous traffic while supporting resource efficiency, QoS, security, and multimedia data exchange.The architecture seeks a convergent gateway solution rather than requiring each IoT subnetwork to maintain its own gateway.
- Network domain: The network domain combines diverse access and core-network technologies, including xSDL, satellite, 3GPP, TISPAN, and LTE-A.This domain provides connectivity to devices, gateways, and application infrastructure.
B. Network-level challenges for IoT
IoT standardization spans multiple initiatives addressing low-power adaptation, routing, constrained application transfer, and machine-to-machine architecture. These efforts remain isolated, motivating broader coordination.
- Current Standardization Efforts: 6LoWPAN adapts IPv6 to low-capacity devices by operating between the IPv6 and MAC layers.It targets compatibility between IPv6 networking and constrained device environments.
- Current Standardization Efforts: ROLL focuses on routing for lower-energy smart-object networks, while CoAP provides web transfer for constrained devices and networks.Both efforts target communication constraints common in IoT deployments.
- Current Standardization Efforts: M2M promotes end-to-end architecture covering sensor integration, naming, addressing, location, QoS, and security.Its scope includes several cross-cutting functions needed for IoT deployments.
- Current Standardization Efforts: These standardization efforts remain isolated rather than forming a unified IoT networking approach.The paper presents their separation as an unresolved issue for deployment.
2) Mobility Management
IoT mobility management must handle numerous mobile objects and changing network conditions. The paper highlights network-wide mobility visibility, adaptive handover, and dynamic flow scheduling as needed capabilities.
- IoT devices frequently operate in diverse mobile scenarios, making efficient mobility management a primary concern.
- Controllers need a network-wide view of IoT-object mobility to manage spatial-temporal requests, adaptive handover, and dynamic flow scheduling.
- Publish/subscribe middleware delivers events asynchronously from source nodes to interested destinations, supporting varied IoT communication patterns.
- Many-to-many IoT communication also creates congestion-control challenges in machine-to-machine scenarios with short packets and short-lived interactions.
4) Communication Protocols
IoT communication protocols must accommodate constrained devices, wireless variability, and security requirements. The described architecture connects heterogeneous IoT devices and gateways through Internet access and DDS publish/subscribe communication.
- TCP imposes connection-setup, congestion-control, buffering, and retransmission overheads that are unsuitable for many IoT scenarios.
- IoT devices require configurable confidentiality, integrity, and privacy protections because wireless channels are vulnerable and devices have limited resources.
- The architecture connects sensors and actuators through local processing, Internet access, gateways, and diverse technologies including Wi-Fi, Ethernet, VPN, USB, Serial, and ZigBee.
A. The Controller Architecture
The controller architecture integrates DDS messaging into SDN to mediate between IoT systems and network services. DDS supports asynchronous, anonymous, many-to-many communication and enables packet handling, forwarding, and flow programming.
- The DDS layer mediates between the IoT system and SDN controller, supporting proactive and reactive flow programming.
- DDS pub/sub provides anonymous, asynchronous, many-to-many communication rather than synchronous point-to-point RESTful interaction.
- The mediation layer implements Packet Handler, Packet Forwarder, and Flow Programming services around controller events and DDS topics.
- DDS messaging transports messages between network-control applications and SDN services while supporting scalability, reliability, flexibility, security, and real-time delivery.
1) Standardization and Open Innovation
The proposed architecture combines standardized SDN and DDS components with programmable wireless networking and virtualized gateways. This combination targets interoperability, mobility, session continuity, multi-homing, and QoS-aware communication.
- SDN and DDS are open, standardized technologies intended to reduce lock-in and support modular network experimentation.
- DDS enables broker-less many-to-many communication, self-addressable and self-routable data, bounded resource use, and message replay after reconnection.
- Virtualized IoT gateways maintain active session continuity and enable sensors to attach dynamically to multiple networks.
- When devices move between networks, the centralized SDN controller modifies OpenFlow switch flow tables to account for mobility.
- DDS supports request-response queues, event-based topics, traffic differentiation, prioritization, selective down-sampling, and DSCP marking for end-to-end QoS.
4) Service Discovery and Interoperability
The architecture combines DDS discovery and traffic-reduction mechanisms with OpenFlow monitoring to improve scalability across IoT gateways and heterogeneous devices.
- DDS discovery lets gateways maintain local object lists containing device locations, descriptions, IP addresses, CoAP URIs, and multicast addresses.
- Gateway-based collection can become a bottleneck when multiple nodes transmit simultaneously, motivating filtering and fusion to reduce unnecessary traffic.
- DDS batching combines multiple small samples into one packet, while Content Filtered Topics deliver only subscriber-requested subsets of a topic.
- OpenFlow complements DDS mechanisms by collecting traffic statistics at configurable granularity and enabling responses to detected network flooding.
6) Context aware network management
DDS multi-channel mechanisms and OpenFlow filtering target high-volume sensor traffic by routing only data that matches application-specified criteria.
- DDS multi-channel supports high-volume sensor streams by dividing data flows across multiple channels.
- Multichannel Data Writers associate multicast addresses with filter expressions so subscribers receive only matching topic data.
- OpenFlow can move keyword-based filtering closer to the source and let a controller application determine filtering decisions dynamically.
7) Solving Security issues
The paper addresses IoT security through DDS security capabilities within a broader DDS–SDN architecture intended to support flexible network orchestration.
- DDS Security Model provides interoperable policies for securing domains, topics, and individual data-object instances.
- DDS partition QoS provides an additional mechanism for creating isolated system segments.
- The proposed data-centric DDS–SDN architecture is presented as enabling agile and flexible IoT network orchestration.
- The study aims to clarify DDS middleware and SDN integration as IoT architectures evolve toward broader technology convergence.