...
In such circumstance, system can send and receive data from several CPEs during one time frame. Such subslots usage significantly decreases latency in highly loaded by subscribers PtMP, however throughput of the whole PtMP decreases too due to higher amount of the service data used and decrease of payload data size. The system use only one subslot in each direction, then minimum average latency for the subscribers (Round-Trip, in case all subscribers are equally active) would equals number of subscriber stations multiple on time frame size and index "2.5". Usage of several subslots decreases latency proportionally to number of subslots. Meanwhile, the more number of subslots is used, the less would be the part of customer payload within the Uplink time frame. Hence with ratio DL/UL 50%, throughput in Uplink part, would be much less than Downlink. It stands to reason Downlink is dedicated only for transmission of data by the base station, however Uplink is shared among all subscriber stations – consequentially, the more number of subscriber stations, the less amount of Uplink is being provided to each subscriber terminal. Accordingly, it is required to ensure symmetrical sector load – then DL/UL ratio should be lower (for example, 45%). Moreover, DL/UL ratio more than 60% with relatively big number of connected subscriber terminals, does not result in increase of throughput Downlink, because decrease of subslots number drastically increases the latency (whereas throughput decrease and each subscriber unit experience output queue size increase), in consequence, ARQ mechanism does not function correctly, because acknowledgement of data requires a lot of time.
...