In the previous blog a brief description of the motivations for LTE-TDD (Time Division Duplexing) were presented. In this occasion, a brief comparison of LTE-FDD and LTE-TDD is presented. Which one do we choose?
The fields compared are:
a) Duplexing: The most significant difference, of course, is the duplexing approach that these systems utilize. In general, FDD spectrum is symmetric by nature due to the way regulations are created. In other words, the government agencies usually have symmetric bandwidth for uplink and downlink for FDD such as 10 MHz in the DL and 10 MHz in the UL.
b) Asymmetry: LTE supports asymmetric bandwidth in the UL and DL. LTE-TDD can be configured to favor one link over the other. Due to the existence of an additional guard time in LTE-TDD, LTE-TDD is slightly less efficient than LTE-FDD, however, it offers more flexibility to serve asymmetric traffic (i.e.: http, ftp, etc.)
c) Hardware Design: Since only transmission or reception is done at a given instant, the designs of the UE and the eNB are much simpler.
d) Channel Estimation: Since TD-LTE uses the same frequency bandwidth for the DL and the UL, the eNB can potentially observe the UL channel conditions to infer the DL channel conditions due to channel reciprocity. Hence, in theory, frequency and the amount of UE feedback about the DL channel conditions could be lower in TD-LTE.
Another fundamental difference is the so-called S-Subframe to make way for switching between DL mode to UL mode. This enables both the UE and the eNB to adjust to either transmitting or receiving data.
a) The 5-ms periodicity: This is used to accommodate delay-sensitive (and symmetric) applications like voice. Also, a 5-ms periodicity makes it feasible for the UE to have compatible operations with WiMAX and TD-SCDMA/UTRA TDD, which also have a frame structure of 5 ms. The downside of using 5-ms periodicity is the additional signaling overhead and the wasting of resources for preparation to switching between the DL and UL.
b) The 10-ms periodicity: This method overcomes the limitations of higher signaling and provides good spectral efficiency. But it can only support asymmetric delay-insensitive applications. Voice applications will be difficult to implement with this kind of frame structure.
The S- subframe is 1 ms in duration and consists of three fields – the DL Pilot Time Slot (DwPTS), the Guard Period (GP) and the UL Pilot Time Slot (UpPTS).