Figure 7.46 shows the UMTS network scenario with multiple interfering sites along
launched network with multiple NodeBs in the area. The inherent feature of UMTS is
that a UE can be associated with multiple NodeBs at a time. Thus the nearby NodeBs or
MEASUREMENTS AND ANALYSIS
76
the neighboring cells that are on active list contribute together to the streaming of data
on downlink path. Hence, the throughput did not drop to 0 even at the poor condition.
But comparatively, LTE 800 test network has only two sites placed such that the cover-
age area of two sites are mostly adjacent and with very less intersection. Thus at few
measurement points, the UE was deprived of coverage and hence the throughput was
ceased to 0.
Figure 7.47 shows the CDF comparison between the DL throughput of Kumpuvaara and
Singerjärvi sites. Range of throughput for Kumpuvaara site is 8.8 Kbps to 10.3 Mbps
while that of Singerjärvi site is 8.8 Kbps to 11.2 Mbps. Mean downlink throughput with
the Kumpuvaara cells as active site is 2.3 Mbps while that of Singerjärvi cells have
mean downlink throughput of 5.1 Mbps. 50
th
percentile for Kumpuvaara site is 1.6
Mbps while that of Singerjärvi site is 5.9 Mbps. Higher percentiles when compared be-
tween Kumpuvaara and Singerjärvi show wider variations while the lower percentiles
almost align to one another. Lower percentile indicates the cell edge conditions are
similar for both of the sites. High deviation in higher percentiles indicates that Sing-
erjärvi site has better throughput within its coverage area. The coverage however for
Kumpuvaara site is higher and thus has more samples. 90% of the samples have
throughput below 5 Mbps while 90% of samples in Singerjärvi have throughput below
9.4 Mbps. The right plot in Figure 7.47 shows the LTE 800 CDF for both sites. Initial
observation indicates that the trend of throughput in LTE 800 is similar to that of UMTS
900. Plot of lower percentiles remain align for both sites till 40%. 90% of samples have
throughput below 20 Mbps approx. for Kumpuvaara while that for Singerjärvi is below
31 Mbps approx. Yet again, the number of samples for Kumpuvaara site is higher than
that of Singerjärvi site with Kumpuvaara site serving at greater distance. The cell edges
or the site is served with poor CQI and resulting in lower application throughput.
Do'stlaringiz bilan baham: