I. Introduction
The parallel operation of voltage source inverters (VSI) in an islanded microgrid (MG) has as a primary goal an even distribution of system load between them, which can be achieved by the droop control method [1]. The frequency deviation inherently induced by the droop method can be solved by diverse restoration strategies. Many solutions (see [2]–[15] to name a few) require exchanging control data between VSIs over a communication network to achieve frequency restoration. The traditional approach to restore the frequency is to apply a centralized structure based on communications where a central unit collects information of all the droop-controlled units, executes a standard PI (proportional-integral) control, and sends back the computed control actions. Even knowing that accurate performance is easily achieved, this approach is sensitive to failures, leading to a single point of failure. To overcome this limitation, recent approaches are inspired in the decentralized control concept, making a different use of the communication channel. And only a few approaches (e.g. [16]–[18]) offer solutions that do not require the exchange of information and therefore they avoid using for control purposes the communication channel of the information and communication technology (ICT) infrastructure that is available in today's MGs [19], [20]. This prevents the degradation that may occurs in distributed control strategies for active power sharing and frequency restoration due to message dropouts, time delays, transmission intervals, quantization, sampling schemes, and traffic scheduling, e.g. [21]–[23].