How Does HSUPA Work?

HSUPA is an improved UL for HSDPA in order to support higher data rates than the ones achieved by R’99 whilst introducing reduced UL latency as well. It uses most of the R’99 features, while employing some new methods of delivering user data from the terminal to Node B. HSUPA is standardised in 3GPP Release 6 and its main characteristics are summarised below:

• Multi-code transmission
  
• Node B scheduling
  
• Option for shorter TTI (TTI = 2ms or TTI = 10ms)
  
• Node B retransmissions (H-ARQ)
  
• Variable spreading factor of minimum SF = 2
  
• Power control
  
• Soft handover

Channel Structure
A new transport channel dedicated to each HSUPA user carrying the user data with HSUPA operation is introduced and denoted as Enhanced Dedicated Channel (E-DCH). Only one E-DCH is dedicated to one HSUPA user and it is mapped onto a pool of newly introduced physical channels (i.e. 1, 2 or 4 E-DCH Dedicated Physical Data Channels - E-DPDCH's and 1 E-DCH Dedicated Physical Control Channel - E-DPCCH). The channel slot structure is shown below:


E-DPDCH carries the HSUPA data and has a variable spreading factor of minimum SF2. E-DPCCH is used for the transmission of the essential information for decoding E-DPDCH and has a fixed SF = 256. It also carries information about retransmissions and power allocation.

E-DPDCH transmission requires simultaneous transmission of the R’99 DPCCH and the E-DPCCH. The first one contains important information about channel estimation and power control, while the other one about decoding.

Three DL physical control channels without carrying any higher layer information are also introduced. These are the E-DCH HARQ Acknowledgement Indicator Channel (E-HICH), the E-DCH Relative Grant Channel (E-RGCH) and the E-DCH Absolute Grant Channel (E-AGCH).

The E-HICH is a dedicated channel with a fixed SF = 128 carrying DL information about retransmissions (i.e. ACK/NACK from the Node B). E-RGCH is also dedicated with SF = 128 carrying the power step up/down scheduling commands from the Node B. Both E-HICH and E-RGCH are transmitted in a single code channel using different signatures. Finally E-AGCH is a common channel with SF = 256 carrying information about the exact amount of power each terminal can use for the E-DPDCH transmission. Each terminal checks whether an E-AGCH is meant for it by recognizing its ID contained in the frame.

Multi-Code Transmission
Adaptive modulation and coding (AMC) is not used in HSUPA, as higher modulation schemes require more energy per bit and the terminals have limited power transmission capability. The solution for achieving higher data rates is found in the use of multi-code transmission, which is more efficient in the UL than AMC.

Multiple (i.e. up to 4) E-DPDCH's can be transmitted for one E-DCH (i.e. one HSUPA user) during one TTI, using code multiplexing. The multiplexed E-DPDCH's can use the same or even different SF's with the allowed combinations being described explicitly in the standards. In order of descending data rates, the allowed combinations for multi-code transmission are 2xSF2 + 2xSF4, 2xSF2, 2xSF4.

For each E-DCH transmission exactly 1 E-DPCCH has to be sent in the same TTI. E-DPCCH and the E-DPDCH'(s) carrying data of the same E-DCH are transmitted using different codes in different branches of QPSK modulation (I-Q multiplexing) leading to transmission of BPSK symbols. The E-DPCCH is always transmitted using I branch.

Node B Based Scheduling
HSUPA scheduling is also shifted to Node B as in HSDPA. However, the scheduling procedure is very similar to the R’99 rather than the HSDPA one. The scheduling algorithm simply tries to keep the cell throughput high by adjusting the E-DPDCH transmit powers and thus the TFC selections. The terminal provides information about its capability of emptying its buffers and the Node B decides whether it can transmit more power or not according to the noise rise limit.

Node B Based Re-Transmissions (H-ARQ)
H-ARQ functionality in HSUPA is very similar to the HSDPA one. The terminal keeps in memory unacknowledged packets and retransmissions take place in case of NACK reception from the Node B. Both IR and CC strategies are supported in HSUPA and the only main difference is that HSUPA HARQ procedure is synchronous. This means that the system knows which HARQ process is used without need for including such information in the data stream.

Option for Shorter TTI
HSUPA allows for the support of two TTI lengths to be chosen; TTI = 2ms and TTI = 10ms. Same as in HSDPA, shorter TTI length (i.e. 2ms) can provide for reduced delays between retransmissions and better utilisation of the resources by reacting faster to the channel conditions. When the channel conditions for a user are changed, the TFC used by the user can be easily changed as well in order to adapt to the new conditions. Furthermore TTI = 2ms supports higher data rates, as above 2Mbps the block size for TTI = 10ms becomes too large and is limited.

However for great number of users, short TTI length requires the transmission of very high number of DL control signals and thus can lead to very high consumption of the Node B power especially, for areas at the cell edge. In HSDPA, the users are time-multiplexed and such problem does not exist. Therefore, the shorter TTI is expected to be used for distances closer to the Node B and as the distance grows, TTI = 10ms will be used.

Email the author