<p>This article addresses the issue of demand-driven cooperative scheduling for distributed generation units (DGUs) under unreliable communication of local power plants. The distributed low-power generation units need to achieve the consensus of the unit states at the point of common coupling. The cooperative scheduling of DGUs is achieved through network communication which is unreliable due to the possible independent denial-of-service (DoS) attacks. Under the circumstances, a dynamic demand-driven communication strategy with switching control approach is provided to efficiently arrange data interactions via the unreliable network subject to attacks. By classifying the disparate interaction modes caused by the transmission mechanism and attacks, the corresponding decay rates of the consensus error system are obtained based on a set of linear matrix inequalities. By introducing the equivalent decay rates for each channel, the stability of the consensus error system is proved in present of DoS attacks. Finally, a DGU simulation example verifies that the designed methods are efficient.</p>