Enhancing End-to-End Determinism and Reliability in 6TiSCH networks with disjoint leaf-based MPLS-like tunnels

TL;DR

This work proposes a cross-layer approach to solve the problem of industrial multi-hop Internet of Things reliability and deterministic behavior, by combining three strategies: the use of the preferred parents (PP) and alternative parents (AP) together with the PRE (Packet Replication and Elimination) technique at the routing level.

Abstract

Industrial multi-hop Internet of Things (IIoT) have strict reliability requirements and they are expected to have deterministic behavior. Reliability is associated with the network's ability to provide the best goodput possible to the destination from the source application, while deterministic behavior implies that the packets must also arrive at the destination before the maximum allowable deadline defined by the application expires. Although a relevant number of proposals have arisen in recent years, none of them achieve both restrictions simultaneously. In this work, we propose a cross-layer approach to solve this problem, by combining three strategies: (i) the use of the preferred parents (PP) and alternative parents (AP) together with the PRE (Packet Replication and Elimination) technique at the routing level; (ii) the use of MPLS tunnels from the leafNode, improving the Data Plane, to control the energy consumption and (iii) the use of the BDPC (Bounded Delay Packet Control) algorithm. The combination of the former strategies show that the behavior of the packet flows improves the end-to-end Packet Delivery Rate of the packets arriving before the deadline by 2.04 times with respect to standard Minimum Scheduling Function reference network while simultaneously increasing the minimum average network lifetime by 1.5 times, with respect to the hop by hop uncontrolled usage of PRE.

Figures (21)

• Figure 1: TSCH slotFrame. The schedule programmed on the slotFrame instructs the node the time to wake up or go to sleep. The cell on coordinate 0x0 is called the MinimalCell and is used for broadcast traffic. Source: AIMARETTO2023100778.
• Figure 2: Location of the Path Computation element (PCE) in a mesh network. The PCE is not really suitable for wireless meshed networks as it cannot react at the same pace as events do develop in the network.
• Figure 3: RPL forms a so called Destination Oriented Directed Acyclyc Graph (DODAG). The Rank helps a node identify its topological position within the network. The smaller the Rank, the closer to the root a node is. Source: AIMARETTO2023100778.
• Figure 4: The strict method to select an alternate parent (AP) implies that the $PP(PP) = PP(AP)$. In this example, $A$ is the PP of $S$. Because $PP(A) = PP(B)$, then $B$ is the AP for $S$.
• Figure 5: Packet Replication and Elimination (PRE). The data packets are copied along the network hop-by-hop, in their path towards their destination. Once the first packet of a given flow arrives to $R$, the rest will be dropped on their arrival.