While the hardware consists of selected and energy-efficient commercial off-the-shelf components, the software defines the heart of WiBACK. Its key components are:
- an MPLS-based traffic forwarding mechanism;
- an IEEE 802.21-inspired network management;
- automatic radio-planning and topology management to set-up and maintain the underlying infrastructure;
- capacity management to distribute available transmission resources and enforce access policies;
- maintaining extensive network monitoring statistics.
MPLS-based traffic forwarding, as it is also used in traditional telecom operator networks, provides virtual tunnels to separate traffic of different traffic classes. Each data packet is associated with an MPLS label, which in turn is associated with a specific traffic class. Even when forwarded between the same radio nodes on the same radio links, a voice packet will experience a different handling than a data packet, and typically be forwarded first. Sophisticated traffic-engineering concepts ensure high network performance, efficient usage of available resources, and provisioning of guaranteed quality of service at the same time.
The use of MPLS also results in reduced demand on the hardware in terms of computing power, thus lower CAPEX and faster message forwarding. Transparent Layer2 forwarding, incl. support for IEEE802.1q VLAN trunking, ensures compatibility with higher-layer protocol such as IPv4 and IPv6. To that end, segments of a WIBACK network form QoS-aware link-local broadcast domains.
IEEE 802.21, an international standard originally developed for inter-technology handovers, was extended by Fraunhofer to support an extended set of control functions. This extension is used to manage the MPLS paths across the network, and to collect and provide network monitoring information. Its technology-independent nature allows WiBACK to integrate any type of radio network technology, provided that an appropriate control interface has been made available. A central component, the Interface Management Function (IMF) provides a uniform and technology agnostic interface to higher layers. MAC adapters located logically below the IMF are responsible for mapping a set of generic primitives onto technology specific features and mechanisms. The higher layer modules on top of the IMF provide functionalities of traditional routing protocols and beyond, such as topology discovery, radio planning, channel assignment, route computation, or monitoring.
The control plane is based on a centralized management approach, where so-called Master nodes manage a set of Slave nodes in their administrative area. Dedicated management entities maintain the resource allocation and forwarding state of their network areas. Multiple Master nodes might be operational within each administrative area in a primary/backup configuration. Contrary to the rather distributed routing protocols such as Open Shortest Path First (OSPF) or Optimized Link State Routing (OLSR), the centralized approach offers the opportunity to perform network wide optimizations when allocating radio resources or when assigning the overall network capacity to best match payload demands.
Automatic topology management, provided by the Topology Management Function (TMF) automatically discovers neighboring nodes, sets up control paths between each node and a Master node, reacts on new nodes as well as node failures, and provides mechanisms for fast re-routing if necessary. The WiBACK Controller uses TMF information to assign radio channels (frequencies) or to select optimal end-to-end paths based on selectable criteria or policies. Monitoring information such as signal quality, link errors or end-to-end QoS violations determine may trigger actions of the self-healing process.
TMF implements a ring-based approach where a Master node first brings up its own radio interfaces and determines the optimal radio configuration. This is computed based on the capabilities of the radio interfaces and the ambient spectrum usage assessed by passive channel utilization analysis. Additionally, TMF may coordinate with external spectrum allocation databases (i.e. TVWS). Once this process is complete, the Master starts sending WiBACK beacons to all its active interfaces to inform adjacent Slave nodes about its availability.
Slave nodes determine their configuration during the bootstrap phase and then switch into a passive beacon scan mode in which they periodically scan all administratively permitted channels for WiBACK beacons sent by a Master node or already associated Slave nodes. Once they detect WiBACK beacons they will attempt to associate with the sending node. If multiple WiBACK beacons have been detected, they will be sorted and associations will be attempted starting with the highest rated sender.
An automated Link Calibration is performed for each newly activated link with the goal to determine, for example, the proper range, modulation and coding (MCS) as well as TxPower settings for each link. Based on this information, the resulting logical link properties, such as capacity and latency are estimated. Those properties serve as the basis of the RMFs constraint-based path computation. In the WiBACK cross-layer design, the TMF may specify the maximum TxPower allowed on a given link, while the respective technology is free to optimally adjust itself to the present channel conditions within the limits set forth by the TMF (i.e. TxPower, MCS, MAC timings).
Logical Resource (i.e. capacity) management is provided by the Resource Management Function (RMF) to ensure that the capacity of a link is never exceeded. Based on configurable policies RMF may adjust its resource allocations to best match end-user demands. Upon association of a new Slave, the TMF re-computes the optimal channel configuration for all available physical radio links in the WiBACK network and may trigger network reorganizations with the goal to optimize the overall network performance (i.e. capacity or latency).
Multi-Layer Technology Independent Monitoring (TIM)
WiBACK maintains extensive network monitoring statistics of its nodes, links and MPLS LSPs, so-called Pipes. This information is continuously examined by the self-management modules to detect possible network problems such as link errors or end-to-end QoS violations. Statistics on the monitoring data are maintained by the Statistics Functions (SF). A Web Interface provides access to this information for administrators. Passive monitoring is used where possible to keep the load caused by monitoring/probing frames at a minimum.