The Sat4Train solution is mainly based on the MLCP, an IP-based smart routing platform able to provide bearer-independent train-to-trackside mobile communication for the railway applications.
It allows traffic prioritization, guaranteeing the reservation of the maximum performances to the railway signalling bi-direction data communication between the onboard the train (i.e. ETCS/EVC) and the control centre (i.e. RBC) .
It is able to integrate heterogeneous wireless technologies such as Satellite Communication and Public Land Mobile Networks (PLMNs) and it can switch to legacy solutions (i.e. GSM-R) in areas already covered by such dedicated networks.
The MLCP technology is ready to integrate also future voice services (i.e. Voice over IP ), once they will be defined and standardized inside the European railway stakeholders community.
The MLCP is composed by two main elements (end-points):
• the on board train platform:
• the back end platform
The on board platform is connected to the ETCS On board Unit (EVC), while the trackside back end is connected to the Radio Block Centre (RBC).
The functional high level architecture of the Sat4Train is depicted in the following figure:
One of the key MLCP technology is constituted by its software, implementing the smart algorithms for both the on-board and the backend equipment.
MLCP provides very efficient end-to-end data transfer and vertical handover by means of MP-TCP (Multipath TCP) which is an extension of TCP protocol to handle multiple paths between two endpoints.
MPTCP splits the transport layer in two sublayers: the upper one responsible of the connection management (establishing connection, reordering packets, etc.), and the lower one gathering a set of simultaneous sub-flows that can be seen as parts composing the same TCP flow.
All these simultaneous parallel data sub-flows can be routed over more independent bearers, associated to different radio access technologies and/or mobile operators, each one offering its services on a best effort basis. The routing policy can be defined to obtain redundancy or, in alternative, load balancing over such paths.
On the basis of all the key QoS parameters, associated to each bearer, gathered and processed in real-time, the MLCP adopts some specific cognitive algorithms to define and tune a predictive model to support the bearers selection and the vertical handover.