Search This Blog

Friday, September 30, 2011

HSDPA fast facts 1



HSDPA fast facts 1

HSDPA basics:
  • Based on shared channel transmission in the time domain (and code domain).
  • 2 ms TTI which enables faster and more frequent adaptation to the varying channel conditions of a time dispersive channel.
  • Rate controlled which is implemented by rapid adjustment of coding rate and modulation scheme.
  • Channel dependent scheduling: since the instantaneous channel conditions vary independently between the active users, it is possible to only schedule users that have good channel conditions for each scheduling moment. This allows for a high network system throughput.
  • For fast rate control and channel adaptation, some L1-2 functionality has been moved closer to the air interface, to the node B.
  • A new sub-layer, Mac-hs, which is responsible for scheduling, rate control and hybrid-ARQ has been introduced and resides in the nodeB

High-Speed Downlink Shared Channel (HS-DSCH)
  • HS-DSCH is the transport channel used to support shared channel transmission and the other HSDPA features mentioned above
  • HS-DSCH consists of a set of 16 channelization codes (each with spreading factor 16) which corresponds to 16 High Speed Physical Downlink Shared Channels (HS-PDSCH)
  • To allow for code resources used for other purposes (R99 DCH) only 15 codes are available for HS-DSCH
  • HS-DSCH resources can be shared in the code domain as well as in the time domain
  • All transmission power that remains after serving all other channels is assigned to HS-DSCH which gives a more or less constant transmission power.
  • Unlike the R99 DCH there is no need to handle issues like DTX or compressed mode with the HS-DSCH. However, actual data transmissions and signaling are indeed suspended during the compressed mode gaps.


Control signaling channels
  • HSDPA downlink control signaling is carried on the High Speed Shared control Channel (HS-SCCH) which is transmitted in parallel to HS-DSCH.
  • ACK-NACK’s for each TTI that the UE has been scheduled in is sent in uplink on High Speed Dedicated Physical Control Channel (HS-DPCCH).
  • Measurements on the downlink channel quality made by the UE are sent in the form of a Channel Quality Indicator (CQI) on the HS-DPCCH.
  • HS-DPCCH has a fixed spreading factor of 256 and 2ms/3-slot structure. First slot is used for HARQ and remaining two for CQI’s.
  • Fractional Dedicated Physical Channel (F-DPCH) containing power control commands for the uplink transmissions are sent in the downlink.
  • Since downlink scheduling takes place in the nodeB, HSDPA does not support macro-diversity or soft handover.

Scheduling:
  • The scheduler in Mac-hs decides what part of the shared code and power resources should be assigned to a user in a certain TTI.
  • Efficient scheduling relies on information about the instantaneous channel conditions which is conveyed by the CQI’s.
  • The CQI value which is received by the node B is directly mapped to a transport block size, modulation scheme and number of channelization codes.
  • Efficient scheduling also relies on buffer status information at the UE.
  • Certain types of data such as RRC signaling is prioritized in the scheduling process.

Rate control
  • The data rate is adjusted for every TTI by selecting the most appropriate modulation, transport block size and channel coding based on the instantaneous channel conditions.

  • Although HSDPA is rate controlled and not power controlled, power can still change due to variations in power requirements for other downlink channels.

HARQ
  • HARQ functionality resides in both physical layer and MAC.
  • HARQ introduces faster retransmissions compared to RLC since there is no signaling between nodeB and RNC and more frequent status reports (every TTI).
  • Due to the shorter TTI in HSDPA the channel conditions are more static during the length of one TTI which improves the accuracy of the status reports.
  • For continuous transmission to a single UE, multiple HARQ-processes can operate in parallel.
  • The number of parallel HARQ processes should match the roundtrip time between the UE and the nodeB.
  • Due to the multiple HARQ processes, the order of the transmitted transport-blocks may become corrupted at the receiver and therefore a mechanism for putting them back in sequence is required before they are passed on to higher layers (RLC)
  • In case a NAK is misinterpreted as an ACK by the transmitting HARQ entity, RLC will detect the error and request the necessary retransmission.
  • For HSDPA, HARQ retransmissions are made in asynchronous and adaptive mode which means that they they can be sent at any time and with any tranport format.
  • A one-bit new data indicator is used to distinguish between retransmissions and new data.
  • ACK/NACK's are always sent at predefined intervals after receiving the transport block. In this way the UE always knows which HARQ process they belong to

Monday, September 26, 2011

Inter-technology carrier aggregation

How cool is this?


Advantages of LTE compared to earlier RAT's



1.
  1. Its possible to do scheduling, link adaptation and interleaving in both the frequency domain as well as in the time domain.
  2. Reduced inter-symbol-interference due to cyclic prefix and lower symbol rate for each sub-carrier. 
  3. Reduced frequency selectivity due to the narrow bandwidth of each sub-carrier which makes is possible to use less complex combining/equalization schemes (MRC instead of MMSE). 
  4. Flexible bandwidth is easy to achieve by altering the number of sub-carriers. 
  5. Efficient equalization for a single carrier bandwidth in excess of 5Mhz is a seriously complex task whereas for a multi carrier system with narrow individual bandwidths the need for equalization is very small.  
  6. Backwards compatibility with earlier technologies is not prioritized if conflicting with LTE performance. 
  7. Maximum throughput in theory for a 20Mhz channel without MIMO is 86,4 Mbps for LTE and 84Mbps for HSPA. For all the above reasons however, average throughput is expected to be 50 % higher for LTE. 
  8. Due to the fact that LTE does not use spreading codes, the scrambling can be directly applied to the code bits which reduces the implementation complexity. However, even with the resources partitioned in the time/frequency domain, scrambling is still needed in LTE to separate the target signal from interference.