Introduction
Historically, network infrastructure has been divided into distinct "silos." Fiber-Wireless (FiWi) networks, comprising high-speed backbones and cellular base stations, are typically managed by large operators. In contrast, Mobile Ad Hoc Networks (MANETs) consist of spontaneous, peer-to-peer connections between devices without reliance on fixed infrastructure.
The lack of coordination between these domains often results in performance bottlenecks. For instance, a cellular base station may be unaware that a mobile user could achieve better throughput via a nearby peer, while a MANET node lacks visibility into the congestion levels of local cellular towers. This paper proposes a federated Software-Defined Networking (SDN) architecture to bridge this gap, enabling dynamic traffic rerouting and fair bandwidth distribution in 5G and IoT environments.
Federated SDN Architecture
The authors introduce a Multi-domain SDN Controller Federation that allows FiWi and MANET domains to exchange monitoring data and control hints. Rather than employing a single, centralized "god-controller"—which would introduce significant complexity and latency—the architecture utilizes a "loose federation."
In this model, separate controllers for FiWi and MANET domains share "hints" regarding network load and user requirements. For example, the FiWi controller can signal congestion at specific base stations, prompting the MANET controller to seek alternative paths through peer-to-peer relays. This allows MANET nodes to act as dynamic relays, redirecting traffic from overloaded Remote Radio Heads (RRHs) to underutilized ones.
Taxonomy of SDN Approaches
To contextualize their contribution, the authors provide a classification system for SDN approaches. This taxonomy evaluates systems based on location distribution, controller integration, management openness, and domain heterogeneity.
As shown in the table above, the proposed "Federated domains" approach is unique in its ability to handle domain heterogeneity and maintain management openness through loosely integrated controllers, making it suitable for complex, multi-operator environments.
Traffic Management and Flow Optimization
The system employs both proactive and reactive management strategies to ensure Quality of Service (QoS):
- Proactive Management: Applications (e.g., high-definition video streams) signal their requirements to the controller before transmission, allowing the network to reserve resources.
- Reactive Management: The network automatically detects "elephant flows"—large, long-lasting data transfers—and triggers rerouting to less-congested paths.
By focusing on elephant flows and ignoring short-lived "mice flows," the architecture significantly reduces signaling overhead. This selective management ensures that the controller is not overwhelmed by the high mobility and frequent connection changes typical of MANET environments.
Conclusion and Future Work
The paper provides a logical proof of concept for load balancing and QoS improvement. Through detailed use-case scenarios, the authors demonstrate that traffic can be successfully redirected through less-loaded paths, reducing latency and packet loss during handovers.
However, real-world deployment faces challenges, including the lack of native SDN/OpenFlow support in consumer mobile hardware and the computational complexity of managing thousands of mobile nodes in real-time. Future research will focus on developing prototypes using ETSI Multi-access Edge Computing (MEC) to optimize high-bandwidth multimedia dispatching, such as smart city surveillance camera streams.
Summary updated