Experimental in-depth study of the dynamics of an indoor industrial low power lossy network

Abstract

An increasing number of industrial applications rely on low power embedded devices because of their flexibility. To work properly, the network has to respect requirements concerning specifically the delay and the reliability. Fortunately, low power, and slow channel hopping MAC help to cope with these requirements. For instance, IEEE802.15.4-TSCH relies on a strict schedule of the transmissions, spread over orthogonal radio channels, to setup a resilient wireless infrastructure. A routing protocol (e.g. RPL) has then to construct energy-efficient routes on top of this link-layer topology. Unfortunately, the radio environment keeps on exhibiting time-varying characteristics, due to e.g. obstacles, and external interference. In a reservation-based stack, the network will have to implement over-provisioning, to cope with small-term variations: additional resources allow the network to operate in the worst situation. Inversely, long-term changes are triggered only when a node/link failure is detected. In this paper, we investigate experimentally the performance stability of a 6TiSCH/ IETF/ RPL stack in collocated deployments. We focus on some key metrics to exhibit the intermittent losses of guarantees (e.g. delivery ratio) under yet static conditions. Our results in large scale testbeds highlight that in the presence of radio oscillations, 6TiSCH introduces frequent network reconfigurations to combat interference and provide high reliability. We perform a multi-layer analysis of the 6TiSCH stack identifying the main sources of instability and proposing solutions to address each one of them. Our performance evaluation highlights the accuracy of our solutions to set up an efficient and reliable network.

SDN-TSCH: Enabling Software Defined Networking for Scheduled Wireless Networks with Traffic Isolation

“…We keep also the default behavior of IEEE 802.15.4-TSCH where EB are transmitted through shared cells, to which all the nodes have to listen to. To reduce the latency and the collisions, we distribute uniformly the shared cells in the slotframe [15].…”

Duocast for Wireless Industrial Networks: an Experimental Study

“…However, such approach is very agressive and the schedule has to be reconstructed from scratch when an inconsistency arises (e.g. an ack is not received) [22].…”

“…This problem is exacerbated by duocast: the transmitter must ush the schedules with the two receivers when a dierence of 6P sequence number is detected. Thus, similarly to [22], we prone a timeout-based maintenance: when a cell is not used for a while, the cell is simply removed silently from the schedule. The timeout in reception is larger than in transmission to guarantee that no packet is lost, i.e.…”