MAC HARQ Assists RLC ARQ

LTE also runs a semi-independent ARQ process at the RLC layer, just above the MAC in the protocol stack, which can be improved with some help from the MAC. “Local NACK,” as it is sometimes known, uses information from the physical layer (PHY) and MAC using the HARQ processes to inform the transmitting RLC entity of missing protocol data units (PDUs) before the data-receiving RLC entity can detect they’re missing. This significantly reduces latency by having the lost RLC PDUs retransmitted much earlier than if the RLC ARQ were used in isolation.

Technical Challenges

The LTE specifications impose constraints on the UE and eNodeB regarding the amount of time they have to complete the HARQ process (Fig. 4). The receiver has three subframes to decode the transmission, check the CRC, and encode the ACK/NACK. Assuming the transmitter sent the data in subframe n, the ACK/NACK must be sent back to the transmitter in subframe n+4.

The transmitter now has three subframes to decode the ACK/NACK returned from the receiver, construct the next transport block(s) based on the ACK/NACK (this is a job for the RLC and MAC), and encode the transport block(s). The next transport block(s) are transmitted on this HARQ process in subframe n+8 in the uplink or potentially earlier for the downlink.

Given that a different HARQ process utilizes each subframe, we can make some assumptions regarding the execution times allowed for each of the processing steps listed above. Assuming only one processing unit for each step is multiplexed in time between all HARQ processes, the computations associated with each step cannot exceed 1 ms.

If one of the steps exceeded 1 ms, it would not be able keep up with the continuous flow of HARQ information each subframe. Drawing some comparisons to W-CDMA/HSPA, the current transmission time interval (TTI) for HSPA is 2 ms, and the TTI for LTE will be 1 ms. The current HSPA UL HARQ turnaround time (transmit, received ACK/NACK, transmit again) is a minimum of 16 ms. This reduces to 8 ms for LTE.

Testing LTE MAC HARQ

UE receiver testing to ensure suitable receiver sensitivity and correct interpretation of the downlink is generally straightforward. In many cases, it can be achieved using only a suitably configured downlink to which the UE can synchronize transmitting some repeatable or pseudorandom bit sequence. Almost no signaling or protocol exchange is necessary.

Testing HARQ, however, requires a full protocol stack, plus some “real” data, and of course channel impairments to simulate poor radio conditions. It is only with all of these factors in place that test will be representative of the real world and repeatable in a laboratory environment.

Testing real applications at data throughput rates representative of the real world will also allow characterization of the user experience. Test equipment would additionally have to be able to provide the variety of channel conditions and react in real time to changes in CQI values provided by the UE. And, it would have to provide some analysis and debug capability in the case of unexpected results.

Summary

Type II HARQ and AMC work together to provide a very adaptive transport mechanism in LTE. AMC tunes the initial HARQ transmission to use a coding rate that will result in approximately the ideal frame error rate from a throughput perspective. Type II HARQ then uses incremental redundancy to add FEC bits in each successive retransmission, reducing the effective code rate until the packet can be decoded correctly. The result, although not perfect, attempts to optimize the overall throughput over wide ranges of dynamically changing channel conditions.

Mark Stambaugh is an R&D system engineer with Agilent Technologies’ Mobile Broadband Division. He holds nine patents and has been with Agilent for the past 21 years, developing one-box cell-phone testers, among other products. He can be reached at stambaugh_temp@agilent.com.

Related Articles:

1. Small-Cell LTE Reference Designs Cut Costs
2. picoChip Will Demo Reference Designs At LTE World Summit
3. Long Term Evolution - What's In It For You?
4. Small-Cell LTE Reference Designs Cut Costs