In this subsection, we introduce several functional split options
that determine the execution of processing in the RAP or in the
cloud-platform and directly influence the required BH data rate.
The discussion is focused on the uplink (UL) since its processing
load dominates the downlink (DL) processing. Detailed
investigations of such splits have also been conducted in [17],
but here we focus more on the opportunities of a flexible split.
By relying on GPPs as opposed to dedicated hardware as used in
the C-RAN concept, and through extensive use of function virtualization,
the envisioned architecture allows us to adapt the
functional split flexibly in time (e.g., according to traffic
demand) and location (e.g., depending on the density of the
deployment). Figure 1 illustrates the principle LTE signal processing
chain of an UL receiver and different options of placing
a functional split. Notice that similar shifts are also possible for
DL processing as considered, e.g., in the context of precoding
for massive MIMO systems in [18].
Subsequently, we discuss these split options and give numerical
results on the required BH data rates per link between one
RAP and the cloud-platform for a simple configuration as specified
in Table 2.
[Table 1] The benefits and signal processing challenges for the centralization
In this subsection, we introduce several functional split optionsthat determine the execution of processing in the RAP or in thecloud-platform and directly influence the required BH data rate.The discussion is focused on the uplink (UL) since its processingload dominates the downlink (DL) processing. Detailedinvestigations of such splits have also been conducted in [17],but here we focus more on the opportunities of a flexible split.By relying on GPPs as opposed to dedicated hardware as used inthe C-RAN concept, and through extensive use of function virtualization,the envisioned architecture allows us to adapt thefunctional split flexibly in time (e.g., according to trafficdemand) and location (e.g., depending on the density of thedeployment). Figure 1 illustrates the principle LTE signal processingchain of an UL receiver and different options of placinga functional split. Notice that similar shifts are also possible forDL processing as considered, e.g., in the context of precodingfor massive MIMO systems in [18].Subsequently, we discuss these split options and give numericalresults on the required BH data rates per link between oneRAP and the cloud-platform for a simple configuration as specifiedin Table 2.[Table 1] The benefits and signal processing challenges for the centralization
การแปล กรุณารอสักครู่..