In this contribution, we consider emerging wireless content delivery networks (CDNs), where multiple (possibly nomadic) clients download large-size files from battery-powered proxy servers via faded links that are composed of multiple slotted orthogonal bearers (e.g., logical subchannels). Since the considered transmit proxy servers are battery-powered mobile routers, a still open basic question deals with searching for optimal energy-allocation (e.g., energy scheduling) policies that efficiently split the available energy over the (faded) bearers. The target is to minimize the resulting (average) download time when constraints on the average available energy per information unit (IU), peak-energy per slot, and minimum energy per bearer (e.g., rate-induced constraints) are simultaneously active. The performance and the robustness of the resulting optimal energy scheduler are tested on the last hop of Rayleigh-faded mesh networks that adopt the so-called ldquodirty paper strategyrdquo for broadcasting multiple traffic flows that are generated by proxy servers equipped with multiple antennas .