In C-RAN (Centralized/Cloud RAN), the more RRHs are covered by one CO (BBU hostelling site), i) the fewer COs are required, and ii) the more BBUs can be centralized at one CO. This allows for the most efficient utilization of BBU resources, thereby significantly reducing CAPEX/OPEX.
4G/LTE Fronthaul Size ,Why should jitter be minimized in CPRI fronthaul? - frequency accuracy
For this reason, the maximum separation distance between RRH and BBU should be secured. This distance is constrained by the timing requirement of Hybrid Automatic Retransmit reQuest (HARQ) protocol used as a retransmission mechanism between UE and eNB in an LTE network.
As seen in Figure 1(a) above, according to this requirement, UE should receive ACK/NACK from eNB in three subframes after sending uplink data, i.e. in the fourth subframe. Otherwise, the UE retransmits the data.
In LTE, uplink and downlink subsframes are typically time-aligned at eNB antenna port as shown in Figure 1(a). So, eNB should complete eNB processing (UL CPRI processing, UL frame decoding, ACK/NACK creation, DL frame creation, DL CPRI processing) within 3 msecs after receiving uplink data from UE in subframe n, and then send downlink ACK/NACK in subframe n+4 back to the UE.
In usual C-RAN, BBU and RRH are located several kms, or even tens of kms, away from each other. So, additional delays like transmission delay via optical fiber, processing time at active equipment in a fronthaul network (e.g. Active WDM, PON, etc.) are caused while data are delivered from the antenna at a cell site to BBU at CO. The sum of these delays and baseband processing time at BBU must be less than 3 msecs.
In order to maintain the timing presented in Figure 1(a), the additional delay caused in the fronthaul network must be compensated somewhere, somehow, for example, by expediting the BBU processing as seen in Figure 1(b).
Base station vendors design BBU to complete the processing and send ACK/NACK usually within 2.75 msecs, intead of 3 msecs, in order to compensate the delay additionally caused in the fronthaul network in C-RAN.
Therefore, about 250 μsecs can be allowed in the fronthaul network. Based on this delay budget, the maximum separation distance between BBU and RRH can be calculated (round-trip transmission latency of 10μs/Km).
In a fronthaul network built with Active WDM, delay components involved in the data transmission after RRH receives data from UE and before it sends ACK/NACK to UE are listed in Table 2 below.
In the table, delay components No. 1 through 3, which are caused at RRH and BBU, must be kept minimized by base station vendors while delay component no. 4 must be kept minimized by fronthaul vendors.
For this reason, the maximum separation distance between RRH and BBU should be secured. This distance is constrained by the timing requirement of Hybrid Automatic Retransmit reQuest (HARQ) protocol used as a retransmission mechanism between UE and eNB in an LTE network.
As seen in Figure 1(a) above, according to this requirement, UE should receive ACK/NACK from eNB in three subframes after sending uplink data, i.e. in the fourth subframe. Otherwise, the UE retransmits the data.
In LTE, uplink and downlink subsframes are typically time-aligned at eNB antenna port as shown in Figure 1(a). So, eNB should complete eNB processing (UL CPRI processing, UL frame decoding, ACK/NACK creation, DL frame creation, DL CPRI processing) within 3 msecs after receiving uplink data from UE in subframe n, and then send downlink ACK/NACK in subframe n+4 back to the UE.
Now, let's consider C-RAN where HARQ is processed between UE and BBU at the CO.
In usual C-RAN, BBU and RRH are located several kms, or even tens of kms, away from each other. So, additional delays like transmission delay via optical fiber, processing time at active equipment in a fronthaul network (e.g. Active WDM, PON, etc.) are caused while data are delivered from the antenna at a cell site to BBU at CO. The sum of these delays and baseband processing time at BBU must be less than 3 msecs.
In order to maintain the timing presented in Figure 1(a), the additional delay caused in the fronthaul network must be compensated somewhere, somehow, for example, by expediting the BBU processing as seen in Figure 1(b).
Base station vendors design BBU to complete the processing and send ACK/NACK usually within 2.75 msecs, intead of 3 msecs, in order to compensate the delay additionally caused in the fronthaul network in C-RAN.
Therefore, about 250 μsecs can be allowed in the fronthaul network. Based on this delay budget, the maximum separation distance between BBU and RRH can be calculated (round-trip transmission latency of 10μs/Km).
In a fronthaul network built with Active WDM, delay components involved in the data transmission after RRH receives data from UE and before it sends ACK/NACK to UE are listed in Table 2 below.
In the table, delay components No. 1 through 3, which are caused at RRH and BBU, must be kept minimized by base station vendors while delay component no. 4 must be kept minimized by fronthaul vendors.
Post a Comment