How Does HSDPA Work?

HSDPA represents an evolution of the WCDMA radio interface, which uses very similar methods to those employed by EDGE (Enhanced Data Rates for GSM Evolution) technology for the GSM radio interface. The fundamental characteristics which enable the increase in the data throughput and capacity with reduced latency are summarised below:

• Time and code multiplexing of the users
  
• Multi-Code transmission
  
• Fixed Spreading Factor (SF = 16)
  
• Shorter TTI = 2ms
  
• No DTX (Discontinuous transmission)for the data channel
  
• Adaptive modulation and coding (AMC) supporting higher order modulation
  
• Node B scheduling and link adaptation
  
• Node B retransmissions (H-ARQ - Hybrid Auotmatic Repeat-Request)
  
• No power control
  
• No soft handover
  

Channel Structure
A new transport channel carrying the user data with HSDPA operation is introduced and denoted as High Speed Downlink Shared Channel (HS-DSCH). This is mapped onto a pool of newly introduced physical channels (i.e. High Speed Physical Downlink Shared channel - HS-PDSCH) to be shared among HSDPA users in a time multiplexed manner. One user only transmits during a single TTI, which has a fixed shorter duration of 3 slots (i.e. 2ms). If code multiplexing is used as well, then multiple users can transmit during the same TTI, using different codes. The HS-PDSCH slot structure is shown below.


Two new physical control channels are also introduced for the handling of associated signalling without carrying information from higher layers. Namely, they are the High Speed Shared Control Channel (HS-SCCH) for the DL and the High Speed Dedicated Physical Control Channel (HS-DPCCH) for the UL.

HS-SCCH has a fixed spreading factor (SF = 128) and carries mainly information essential for demodulation and retransmissions. One HS-SCCH per TTI is required for each active HSDPA user. HS-DPCCH has a fixed spreading factor (SF = 256) and carries mainly channel quality information and ACK/NACK for retransmissions.

An HS-PDSCH corresponds to one channelling code of fixed spreading factor SF=16, and thus there are maximum 15 such codes available for HS-DSCH transmission under one code tree, as at least 1 such code is needed for UMTS and HSDPA control channels. It is noted here that the same code tree is used for both UMTS and HSDPA traffic when a single carrier deployment is used, to do otherwise the use of multiple scrambling codes per cell would be required. The use of a separate carrier for HSDPA would add another code tree and thus increase the capacity of the system.

Multi-code transmissions are also supported in HSDPA, which means that the user equipment can be assigned multiple codes during one TTI, depending on its capability. This allows users to transmit large amounts of data in one TTI leading to a better resource utilisation which increases system throughput. Throughput is also increased by not supporting DTX for HS-PDSCH and users are scheduled only if they have data to transmit.

AMC (Adaptive Modulation and Coding)
An HS-PDSCH may use QPSK or 16-QAM modulation symbols, and several channel coding rates depending on the channel conditions, as reported by the terminal in the HS-DPCCH frames. Thus the system adapts to the channel conditions and can achieve higher data rates when the conditions are favourable. This procedure is denoted as link adaptation and has similar effects as power control without need for power control overhead. Each modulation scheme and coding rate combination is denoted as Transport Format Combination (TFC) or Modulation and Coding Scheme (MCS).

Node B Based Fast Scheduling and Link Adaptation
The scheduling and link adaptation in HSPDA are performed in the Node B unlike R’99, where scheduling is done in the RNC. Node B estimates the channel quality for each active user based on the feedback provided by them in the HS-DPCCH frame. Afterwards the user to be served first depends on the scheduling algorithm and the user prioritisation.

In general, there are three principal scheduling algorithms depending on the equipment (the algorithms used to implement these can take many different forms) and these are:

• C/I based (Max C/I) – serves in every TTI the user with the largest C/I
  
• Round Robin (RR) – users are served in a cyclic order ignoring the channel quality conditions
  
• Proportional Fair (PF) – serves users with the largest relative channel quality (i.e. the channel quality conditions are weighted by the user throughput)

Node B scheduling in combination with the shorter TTI provides with faster scheduling and better resource utilisation, as the network reacts faster to the channel conditions and higher rates can be achieved for users with good channel conditions.

Node B Based Retransmissions (H-ARQ)
Apart from scheduling, retransmission handling is also shifted from the RNC to Node B in HSDPA and thus from RLC to the physical layer. The packets sent to a user are kept in the Node B buffer until an ACK is received from the terminal carried in the HS-DPCCH frame. The RNC will retransmit only in case that the physical operation fails. This makes retransmissions faster, reducing the latency but also provides with combining gain, as the terminal keeps the energy of both transmissions.

The retransmission protocol selected in HSDPA is the Stop and Wait (SAW) due to the simplicity of this form of ARQ. The Hybrid ARQ (H-ARQ) technique is fundamentally different from WCDMA retransmissions because the UE decoder combines the soft information of multiple transmissions of transport blocks at bit level imposing significant memory requirements on the UE. Two different Hybrid ARQ strategies are generally used:

• Chase Combining (CC) – the basic idea of the CC strategy is to transmit identical versions of the erroneously detected data packet and then use maximal ratio combining effectively providing with time diversity and soft combining gain.
  
• Incremental Redundancy (IR) – additional redundant information is incrementally transmitted if the decoding fails on the first attempt.

Email the author