Local or Edge/Cloud Processing for Data Freshness

Several actuation choices in the Internet of things (IoT) require state information to be up-to-date. In this context, the choice between local processing and offloading to a more powerful remote server can have a significant impact on data freshness. Local processing is indeed typically less reliable due to limited computing power and may require multiple attempts, which can result in data becoming stale. Conversely, remote (i.e., edge/cloud) processing can be more reliable, yet entail longer response times due to data transmission to an external server, possibly performing processor sharing with other tasks. The optimal balance depends on the specific system conditions, especially on the congestion at the remote server, and is in general non-trivial. We explore this tradeoff through a mathematical model, and argue about the derivation of a freshness-efficient local vs. remote split of requests, providing insights on the role played by reliability and latency in selecting the optimal strategy.