In LTE Scheduling means at what time, and on which subcarriers, information for/of which MS is transmitted. Furthermore, it also involves the choice of the transport format, i.e., transport block size, modulation and coding scheme, and the multiple-antenna scheme. Under normal circumstances, the scheduling decision is transmitted afresh in every subframe, since it can change from subframe to subframe. However, in the case of voice calls (and other applications that have a low data rate but continuous data), semipersistent scheduling is used. Essentially, the Base Station tells the MS that (until further notice) it has the same resources allocated to it in every $n^{th}$ frame; after that message, no further scheduling information needs to be transmitted to that Mobile station, thus reducing the overhead. The principle of operation of scheduling in HARQ is illustrated in Figure 15.

All scheduling decisions are made by the serving Base Station (though that BS might use information it received from the Mobile stations and from other BSs, in order to provide a better quality). This is true for both the downlink and the uplink. Note, however, that an MS that has multiple radio bearers to transmit for the uplink decides by itself which one to transmit on the resources assigned to it by the Base Station. The standard does not define how the scheduling is to be done. In general, the scheduler will try to exploit the multiuser gain as much as possible, but also the interference to other cells, as well as the latency (due to backlog) for each Mobile station must be taken into account.