Professional Documents
Culture Documents
ZTE University
Content
MAC-d
MAC-es / MAC-c/sh
MAC-hs
MAC-e
MAC - es/e
HARQ
Associated Scheduling
Downlink Signalling
(E- AGCH / E - RGCH(s) )
Both uplink logical channel DCCH and DTCH can be mapped to E-DCH
HSUPA new physical channel
Five physical channels are added in radio interface to
support rapid re-transmission, soft combination and Node
B distributing scheduling in physical, these five physical
channels all support 10ms TTI and 2ms TTI.
E-DPDCH: E-DCH Dedicated Physical Data Channel
(uplink).
E-DPCCH: E-DCH Dedicated Physical Control Channel
(uplink).
E-HICH: E-DCH HARQ acknowledgement Indicator
Channel (downlink).
E-AGCH: E-DCH Absolute Grant Channel (downlink).
E-RGCH: E-DCH Relevant Grant Channel (downlink).
Uplink dedicated physical channel E-DPDCH
and E-DPCCH
E-DPDCH is a uplink dedicated physical data channel to bear the data of E-DCH
E-AGCH is a downlink common physical channel to bear the absolute grant of E-DCH
1 subframe = 2 ms
1 radio frame, T f = 10 ms
Downlink dedicated physical channel E-RGCH
E-RGCH is a Downlink dedicated physical channel to bear the relative grant of E-DCH
Adopt fixed SF=128.
A relative grant can be transmitted in 3, 12 or 15 continuous time slots.
3 and 12 time slots are corresponded to 2ms TTI and 10ms TTI in
serving E-DCH cell, 15 time slots is used in non-serving E-DCH cell.
1 subframe = 2 ms
1 radio frame, Tf = 10 ms
Downlink dedicated physical channel E-HICH
E-HICH is a Downlink dedicated physical channel to bear the HARQ acknowledgement Indicator of E-DCH
1 subframe = 2 ms
1 radio frame, T f = 10 ms
Basic Concept
E-DCH active set:
The cell where the UE receives AGs (absolute grants). One UE has only one
serving E-DCH cell.
Iur/Iub FP
E-HICH (ACK/NACKs)
E-RGCH (relative grants)
E-DCH Serving
serving cell (ChCode, signature -> UE)
Non-serving
Cell RLS
E-AGCH
(Absolute Grants,
"E-RNTI" -> UE) E-DPDCH
E-DPCCH
UE
MAC-e/
HSUPA working
MAC-es Only one cell is in charge of E-DCH
principle under MAC-d scheduling under the state of non-soft
soft handover DTCHs handover, that is E-DCH serving cell.
HSUPA basic working process (1)
Establishing process of dedicated channel E-DCH
UE sends service request to network side as the same process of
R99/R4.
Upon receiving RAB establishment request, the SRNC determines to
select uplink E-DCH according to service attributes and sends the RL
SETUP message to NodeB. The RL SETUP message indicates which
RL is the E-DCH RL and which RL is the serving E-DCH RL.
After establishing RL, the NodeB sends RL SETUP response to RNC.
The response message contains the E-AGCH/E-RGCH/E-HICH
scramble and channelization code, and E-RGCH/E-HICH signature
sequence. If the RLs contain the serving RL, the NodeB allocates E-
RNTI to the UE. The response message also contains the E-RNTI.
The RNC sends UE the RB SETUP message, carrying E-RNTI,
mapping relationship between RB and Mac-d Flow, E-TFCS, Mac-d
Flow, E-AGCH/E-RGCH/E-HICH code resource and signature
information.
HSUPA basic working process (2)
RV Selection 11520
Max. Transmission
E-DCH Codes X Data Rate
TTI Block Size (bit)
Category Spreading (10ms/2ms)
(10ms/2ms)
Category 1 1×SF4 10ms 7110 0.71Mbps
1.45Mbps/1.42
Category 2 2×SF4 10ms, 2ms 14484/2798
Mbps
Category 3 2×SF4 10ms 14484 1.45Mbps
2Mbps/2.9Mb
Category 4 2×SF2 10ms, 2ms 20000/5772
ps
Category 5 2×SF2 10ms 20000 2Mbps
2×SF2+ 2Mbps/5.76Mb
Category 6 10ms, 2ms 20000/11484
2×SF4 ps
Remark: Data Rate = (Max. Transmission Block Size + 36bit CRC) / TTI
Content
Node B uses different ways to combine the multiple retransmissions of a single packet,
decrease the receive Ec/No of each transmission. With HARQ, HSUPA can effectively
increase data transmission rate and shorten transmission delay.
Retransmission comparing between HSUPA
HARQ and R99 RLC
The fast HARQ may cause higher target block error rate (BLER) in the first transmission,
because it enables shorter delay in resending the packets that are not correctly received
previously compared with RLC retransmission.
The higher BLER target can reduce the transmit power that is required by the UE in
transferring the data of certain rate. Therefore, for the same cell load, the fast HARQ can
increase the cell capacity.
When the data rate is fixed, the energy decreasing of each bit can improve the coverage.
Uplink HARQ Soft Handoff
In CDMA system, the software handoff gain
is generated when a Node B receives
packets correctly, while another Node B
cannot decode.
In the soft handoff of HSUPA HARQ, one
Node B sends acknowledgement (ACK) to
the UE, while the other one sends negative
acknowledgement (NACK). In this case, the
network already receives the packet, and the
UE should not sent the same packet.
Accordingly, the HARQ process of the Node
B that receives packet incorrectly can
resume from the reception failure. The RNC
must ensure the sequence of the packet and
send it to the upper layer, and performs
selective combination of the packets
received from different Node Bs.
In the process of SHO, UE can perform retransmission only when all Node Bs in the active set
can’t decode correctly, otherwise, UE will not perform retransmission if there is one ACK.
SHO enables link diversity gain, improves throughput effectively, and reduces retransmission
times.
Fast Packet Scheduling
To R99/R4/R5, packet
scheduling is based on
RNC.
In HSUPA, the
scheduling is located in
Node B, and the
scheduling period can
be 2ms, to implement
fast scheduling strategy,
thus, the uplink air
interface capacity can be
effectively utilized.
The packet scheduling controlled by RNC will cause some delay, so the change of current
channel can not be reflected quickly, thus, the fast link self-adaptation and fast packet
scheduling can not be performed.
In HSUPA, the packet scheduling entity of Node B can directly use real-time measurement
information of physical layer, inner statistic information and the information reported by UE to
perform scheduling, thus, to reduce the system transmission delay by utilizing the situation of
channel and fading attribute of different users.
Fast Uplink Packet Scheduling
Uplink and downlink radio packet scheduling have the different
complication of resource controlling.
In downlink: Node B can easily get the information of buffer occupation of each
service streaming, and the downlink power is central controlled, so Node B can
allocate and free the power resource accurately, thus to control the system
load efficiently and ensure the QoS of each service.
In Uplink: the time delay and uncertainty are existed in uplink service streaming
cause the information of it need to be reported to Node B by radio channel. In
uplink, each UE has its own power source, because of the time variability of
radio channel and the limitation of code resource, the accurate control of uplink
system resource (total uplink power) became more complicated, so as to the
uplink packet scheduling.
Radio Channel
queue of
user 1
Service Resource
queue of
…… allocation
user 1
In order to confirm the uplink transmission rate and transmitting power, some
information of user must be known like the situation of serving streaming and
available power, etc. which is named SI (Scheduling Information).
SI is periodically reported in HSUPA, and the period of report is integral times
of TTI, SI is multiplexed with packet data.
Node B allocates uplink resource to user according to the SI information
multiplexed from received data, that is, Node B sends resource indication of
scheduling Absolute Grant by downlink AGCH channel.
Scheduling Resource Compensation
Mechanism
The shortage of SI transmission
The error of SI transmission is unavoidable and the report is
unreliable cause of being transmitted by the radio channel.
Node B adopts periodical scheduling, but the SI may be not
accurately reflect the UE situation at real time.
While the situation of radio channel is bad, the SI will not be
received correctly before it has been multiple retransmitted.
That causes time delay.
Compensation Mechanism
The obvious reduction in scheduling period enables the uplink air interface
capacity to be better controlled dynamically, and the resource of air interface
will be effectively utilized.
The potential advantage is that the running target of uplink load can be more
approached to the maximal level of load, but it will not improve the probability
of overload, thus, the probability of uplink noise rising is lower than that caused
RNC scheduling.
Effect of Fast Packet Scheduling on Maximum
Expected Load
The probability of uplink load
target is much more near that of
maximum load limit when the
marginal load area becomes
smaller (The Prx_target is more
closer to the Prx_threshold. The
Prx_threshold is the overload
threshold).
The Node B-based scheduling
requires the information related
to UE uplink transmission
demand. The UE should be
informed of the current data rate
allowed through fast signaling
within specified time.
To RNC, these operations are
implemented through RRC
signaling, the response speed of
which is restricted obviously.
2ms TTI
E-DCH Max. Min. Max. transmission Max. transmission
type E-DCH Spread TTI type block size block size
channels Factor (10ms TTI) (2ms TTI)
Category 1 1 SF4 10 ms TTI only 7110 -
As shown in the prior table, the 12.2k, 64k service of R99 almost have
the same coverage, the radius of dense urban is about 450m, the
distance between two sites is about 780m. The low speed HSUPA
service (mode1) can have the same coverage as 12.2k.
In mode2, it is very hard to bear 480kbps, because there is only one
SF code. And the radius of coverage is shrunk to about 240m.
In mode3, for the bearing speed is improved and the SF code is
increased to 2, the radius of coverage can be kept in about 240m.
For the high speed uplink service (mode4), the bearing speed is
improved to 2Mbps with 2 SF codes, And the radius of coverage is
shrunk to about 190m.
Link Budget Comparison of HSUPA/R99 in
Co-frequency Networking Mode
R99 HSUPA
Maximal
124.936 125.068 126.338 118.237 118.137
allowed 127.1382 115.0773
4 2 2 6 6
path loss
HSUPA network construction strategy: Cover the whole network with relatively low rate.
Focus on the coverage of dense urban areas, to ensure the high rate of uplink
transmission in hotspot areas. It is recommended that for the planning the initial stage, it is
advisable to refer to HSDPA coverage range, conduct key coverage in hotspot areas, and
just conduct the low speed coverage in the edge areas of common cell.
In HSUPA planning, under the conditions that the same overload probability is ensured,
and RRM control algorithm is further optimized, in order to enhance uplink throughput, in
setting uplink interference margin, it is recommended not to use the common R99 uplink
3dB as the noise raise threshold, and the value 5 ~ 6dB is recommended.
In the co-frequency planning of HSUPA and R99, if the interference margin threshold is
raised, it will cause the previous R99 uplink planned budget radius to shrink. As to how
much it will shrink specifically, it is necessary to conduct subsequent and further study.
From link budget analysis, it shrinks by about 10 ~ 20%. In inter-frequency networking,
there is basically no influence on R99 uplink planning.
Via the link budget, it can be known that HSUPA 64k service budget radius is slightly
smaller than R99 12.2k and 64k service radiuses under the inter-frequency HSUPA
networking conditions; HSUPA 64k service budget radius is slightly bigger than R99 12.2k
and 64k service radiuses under the co-frequency networking conditions. Analyzed in a
comprehensive manner, the low speed HSUPA service can implement the same coverage
with the planned R99 range.
HSUPA theoretic data rate
HSUPA Enhanced Uplink Dedicated Physical Channel E-DPDCH supports SF=2~256,
and multiple channels. Adopts QPSK modulation and coding rate 1/3 (turbo). If the
coding rate is 1, then the maximal transmission capacity of Enhanced Uplink Dedicated
Physical Channel is up to 5.76Mbit/s by data rate matching arithmetic。
CC(480kbps)
NC(480kbps)
400 CC(144kbps)
NC(144kbps)
Soft Combining
Throughput [kbps]
Influence of HARQ 300
200
100
0
-22 -20 -18 -16 -14 -12 -10 -8 -6 -4
received Ec/No [dB]
In the HARQ protocol, it is necessary to retransmit the data that can not be correctly received. And
it is possible to generate the gain in throughput for the UE whose transmission power is limited by
using Soft Combining, and the retransmission times can be effectively reduced.
The algorithm which adopts the soft combining brings much more gain than that which does not
adopt the soft combining. This is because that the soft combing fully utilizes the bit information of
the previous transmission. Therefore, the gain is more remarkable when the Ec/No is relatively low.
HSUPA data rate transmission efficiency (3)
Throughput comparison under PA 3km/h & no soft handover
CC, SHO (2 links), PA 3kmph, 144kbps, 0dB imb
150
Throughput [kbps]
100
Macro Diversity
Influence
50
w/ div
w/o div
0
-22 -20 -18 -16 -14 -12 -10 -8 -6 -4 -2
Total received Ec/No in the active set [dB]
When macro diversity technology is adopted, the UE will conduct the retransmission only
when no Node B in all the activated sets can correctly decode the data. Otherwise, the UE
will not execute the retransmission as long as there is one ACK.
Adopting macro diversity brings remarkable performance gain, effectively improves
throughput, and reduces the retransmission times.
HSUPA cell performance analysis
HSUPA cell performance analysis is evaluated in the following aspects: The
average throughput of cell, average throughput of a single user, UE
scheduling fairness, service delay and RoT overload rate
HSUPA cell throughput demonstrates the function relationship between the average RoT
of the cell and the system throughput. Besides being influenced by such factors as the UE
quantity, the geographical position where the UE is located and the wireless propagation
condition, UE data transmission request, and the uplink interferences of the cells around it,
it is also influenced by different wireless and packet scheduling polices and scheduling
algorithm.
The fairness curve refers to the density accumulation distribution function of the
normalization of user throughput by the average throughput of each user, and it reflects
the scheduling fairness.
The service delay includes the packet call delay and packet delay. The packet call delay
refers to the time period between two consecutive packet transmission requests. For the
FTP user, the packet call delay refers to the time the FTP uploads the file. The packet
delay refers to the time necessary for the reception of the packet message at Node B.
RoT overload ratio reflects the ratio of RoT when it exceeds a certain designated RoT
threshold value
Influence of 2 ms TTI on system performance
Average Packet Delay for FTP Users
Comparison of 2ms and 10ms Cell Throughput- Mixed Channel Packet Delay for FTP Users
35
2ms vs. 10ms, mixed channel, PF
30
20
30
15
20 10
5
10
0
0 3.5 4 4.5 5 5.5 6 6.5
Average RoT [dB]
3 3.5 4 4.5 5 5.5 6 6.5
TTI=2ms TTI=10ms
Avg. RoT (dB)
Average Packet Delay for Video Users
In HSUPA, the shorter frame structure Packet Delay for Video Users
1.8
whose TTI is 2ms is adopted. The shorter 1.6
We analyze, via simulation, the influence on the average throughput of the cell
in the system when there has soft handover (includes softer handover) and
when there is no soft handover. It can be seen that due to the adoption of soft
handover technology, the link diversity gain is generated, and the cell
throughput is improved by at least 120 kpbs.
Content
WCDMA
GSM GPRS HSDPA HSUPA LTE
R99
EDGE
HSPA+
CDMA2000
1X EV-DV
UMTS Technology Evolution Roadmap
Year
2002-3 2003-4 2005-6 2007-9 After 2009
64-144kbps 64-384kbps 384kbps-4Mbps 384kbps-7Mbps 20-50Mbps
Downlink
Throughput
3G NGMN
3G+HSDPA HSDPA/HSUP (LTE,…)
GSM 3G Downlink A Broadband radio
R99 IP based wideband
GPRS/EDGE Enhanced Downlink/Uplink
Peer to Peer
Enhanced
NGMN
Optimized UMTS
Enhanced UMTS
3G
GSM(GPRS/EDGE)