Nov 17, 2017
Jolley Hall, Room 309
"Real-Time Reliable Middleware for Industrial Internet-of-Things"
Adviser: Chris Gill
This proposal aims to contribute to the real-time and fault-tolerant middleware research that applies to Industrial Internet of Things (IIoT). IIoT requires event processing and delivery, subject to application-specific combinations of timing and delivery requirements. Such a middleware must both differentiate latency according to the applications' requirement, and enforce cyber-physical events' timing constraints (such as event freshness and relative consistency of each other, at the time of use). The middleware must also enforce reliable delivery such as at-least-once, for critical IIoT applications. It is nontrivial to efficiently differentiate such heterogeneous IIoT timing/reliability requirements to fit each application, especially when common fault-tolerant mechanisms tend to slow down the system (such as those using checkpointing and replication).
This proposal presents a new real-time reliable IIoT middleware, with the following three contributions: (1) Real-time cyber-physical event processing, implemented within the CORBA-compliant TAO event service. The resulting middleware both differentiates application-specific latency requirements and enforces cyber-physical timing constraints, by prioritizing, sharing, and shedding event processing. (2) Real-time reliable IIoT event service, which provides a configurable checkpointing and replication framework that integrates suitable replication types to fit each application's unique timing and delivery requirements. (3) A new real-time efficient replication type, which reduces both resource redundancy (compared to active/semi-active replication types) and the incurred latency (which is typical of passive replication types). The new replication is to be integrated as part of the proposed real-time reliable IIoT middleware.