Requests of resources of uplink flows towards a base station (BS) and/or downlink flows towards end nodes are computed by each requesting node on an end-to-end basis. Each requesting node issues a cumulative request given by summing up the same request for each link separating the node from the BS plus each link separating the BS from the destination node, enabling the centralized node to perform connection based scheduling. The BS in response to all cumulative requests grants uplink and/or downlink resources for each link. The profile of grants is equal to the profile of requests when the amount of resources requested is below the maximum net throughput on MAC layer. When not below the maximum net through-put on MAC layer, the ideal profile of grants is normalized with respect to a ratio between the maximum net throughput on MAC layer and the amount of the overall requested resources.