interval. This transition causes non-negligible amount of
energy to be spent. When the timer is increased to 20s,
this transition no longer occurs and the average current is
decreased. Even further savings should be possible when increasing
the burst interval and optimizing the cDRX profile.
We expect similar results to apply for the video streaming
case except that the relative energy savings will be smaller
because the baseline current is higher due to active display
and more computational work.
As for the signaling load, there can be two kind of state
transitions as there are only two states in LTE RRC. Each
transition causes a certain amount of signaling within the
network. We note that using the configuration that delivers
largest energy savings, namely cDRX enabled and inactivity
timer longer than the burst interval, there is no increase in
signaling traffic due to state transitions. The cDRX mechanism
itself does not cause any extra signaling load to the
RAN compared to a case where a phone is in connected
mode without DRX.
In order to compare different technologies, we measured
also the case when streaming over 3G with the same phone
(HSPA in Figure 7). There is a notable difference in the average
current when streaming directly from the server in favor
of HSPA but the difference becomes negligible if cDRX is
activated. The figure includes also the current draw caused
by listening FM radio with the same smartphone. Although
the energy savings can be cut by more than half with traffic
shaping and cDRX, the resulting current is still double the
plain old radio. We measured similarly the FM radio current
of N900 and the result was the same 100 mA than with
the LTE phone. However, the lowest average currents we
measured for audio streaming over 3G are only marginally
higher than the average FM radio current.
interval. This transition causes non-negligible amount ofenergy to be spent. When the timer is increased to 20s,this transition no longer occurs and the average current isdecreased. Even further savings should be possible when increasingthe burst interval and optimizing the cDRX profile.We expect similar results to apply for the video streamingcase except that the relative energy savings will be smallerbecause the baseline current is higher due to active displayand more computational work.As for the signaling load, there can be two kind of statetransitions as there are only two states in LTE RRC. Eachtransition causes a certain amount of signaling within thenetwork. We note that using the configuration that deliverslargest energy savings, namely cDRX enabled and inactivitytimer longer than the burst interval, there is no increase insignaling traffic due to state transitions. The cDRX mechanismitself does not cause any extra signaling load to theRAN compared to a case where a phone is in connectedmode without DRX.In order to compare different technologies, we measuredalso the case when streaming over 3G with the same phone(HSPA in Figure 7). There is a notable difference in the averagecurrent when streaming directly from the server in favorof HSPA but the difference becomes negligible if cDRX isactivated. The figure includes also the current draw causedby listening FM radio with the same smartphone. Althoughthe energy savings can be cut by more than half with trafficshaping and cDRX, the resulting current is still double theplain old radio. We measured similarly the FM radio currentof N900 and the result was the same 100 mA than withthe LTE phone. However, the lowest average currents wemeasured for audio streaming over 3G are only marginallyhigher than the average FM radio current.
การแปล กรุณารอสักครู่..