Is Link-Layer Anycast Scheduling Relevant for IEEE 802.15.4-TSCH Networks?

With the wide adoption of low-power wireless transmissions, industrial networks have started to incorporate wireless devices in their communication infrastructure. Anycast is a link-layer technique to improve the reliability when using lossy links. Several receivers are associated to a single transmission, so that a transmission is considered erroneous when none of the receivers was able to decode and acknowledge it. Appropriately exploited by the routing layer, we can also increase the fault-tolerance. However, most of existing anycast schemes have been evaluated by simulations. Besides, most evaluation models assume that packet drops are independent events, which may not be the case for packet drops due to e.g. external interference. Here, we use a large dataset obtained from an indoor testbed to assess the gain of using anycast in real conditions. We also propose a strategy to select the set of forwarding nodes: they must increase the reliability by maintaining packet losses as independent as possible. Based on our experimental dataset, we demonstrate the efficiency of anycast, provided that next hop selection follows a specific set of rules at the routing layer.