1: \begin{abstract}

2: In this paper, we design, analyze the convergence properties and address the implementation aspects of \textit{AFAFed}. This is a novel \textit{A}synchronous \textit{F}air \textit{A}daptive \textit{Fed}erated learning framework for stream-oriented IoT application environments, which are featured by time-varying operating conditions, heterogeneous resource-limited devices (i.e., coworkers), non-i.i.d. local training data and unreliable communication links. The \textit{key new} of \textit{AFAFed} is the synergic co-design of: (i) two sets of adaptively tuned tolerance thresholds and fairness coefficients at the coworkers and central server, respectively; and, (ii) a distributed adaptive mechanism, which allows each coworker to adaptively tune own communication rate. The convergence properties of \textit{AFAFed} under (possibly) non-convex loss functions is guaranteed by a set of \textit{new} analytical bounds, which formally unveil the impact on the resulting \textit{AFAFed} convergence rate of a number of Federated Learning (FL) parameters, like, first and second moments of the per-coworker number of consecutive model updates, data skewness, communication packet-loss probability, and maximum/minimum values of the (adaptively tuned) mixing coefficient used for model aggregation. 

3: \end{abstract}

4: