Transmission over multiple component carriers in LTE-A uplink

TLDR: Results show that when a linear receiver is used in the base station, mixing techniques can increase spectral efficiency, thus reducing the performance gap to the no bundling case, which is the most expensive solution in terms of feedback signaling.

Abstract: Long Term Evolution - Advanced systems are currently being standardized by 3GPP and aim at very high peak data rates of 1 Gb/s in the downlink and 500 Mb/s in the uplink. Those ambitious targets can only be achieved by using advanced MIMO antenna techniques as well as wide spectrum allocation, up to 100 MHz. A multiple component carrier structure has been agreed on in the 3GPP Work Item as a solution to extend the 18 MHz bandwidth of the previous LTE Release 8 up to 100 MHz. The multiple access schemes on both uplink and downlink now have to be adapted to the new spectrum configuration. Furthermore, in the link adaptation design the transmission over multiple CCs would reasonably lead to an increase of the feedback overhead. Bundling of the spatial or frequency parameters can keep the overhead low at the cost of lower throughput. In this article, we consider as a study case the LTE-A uplink, where NxDFT-spread- OFDM has been selected as the multiple access scheme. The validity of this scheme for the uplink is evaluated in terms of cubic metric, which is an indicator of the power de-rating needed at the transmitter to avoid intermodulation distortion. Furthermore, the impact of bundling the link adaptation parameters on the link performance is discussed considering both linear and turbo successive interference cancellation (SIC) receivers. Two codeword mixing stategies in the frequency and spatial domains are also proposed to boost the performance when the bundling is made per antenna or per CC, respectively. Results show that when a linear receiver is used in the base station, mixing techniques can increase spectral efficiency, thus reducing the performance gap to the no bundling case, which is the most expensive solution in terms of feedback signaling. However, when a turbo SIC receiver is used, only mixing over CCs is a valid option to achieve link performance gain.

Figures

Fig. 1. Multiple CC spectrum structure.

Fig. 3. CM performance of NxDFT-s-OFDM and OFDM.

Fig. 2. Signal generation for (a) DFT-s-OFDM, (b) NxDFT-s-OFDM, with the option of clustered allocation of the RBs (c).

Fig. 5. Spectral efficiency performance of the bundling options with (a) 2 CCs and linear receiver, (b) different number of CCs, turbo SIC receiver with (c) low speed and (d) high speed.

TABLE I FEEDBACK OVERHEAD FOR LINK ADAPTATION (BITS PER FRAME)

Frequently asked questions

Q1. What is the main selling point of the SC-FDM technology?

A low CM property translates to higher power efficiency and therefore longer operation time; furthermore, it can improve coverage since the lower power de-rating compared to the use of OFDM allows users at the cell edge to transmit with relatively higher power. 

Q2. What have the authors contributed in "Transmission over multiple component carriers in lte-a uplink" ?

Long Term Evolution-Advanced ( LTE-A ) systems are currently being standardized by the 3rd Generation Partnership Project ( 3GPP ) and aim at very high peak data rates of 1 Gbits/s in the downlink and 500 Mbits/s in the uplink. In this article, the authors conside as a study case LTE-A uplink, where the NxDFT-spread-OFDM ( NxDFT-s-OFDM ) has been selected as multiple access scheme. Furthermore, in the link adaptation design the transmission over multiple CCs would reasonably lead to an increase of the feedback overhead. Furthermore, the impact of the bundling of the link adaptation parameters on the link performance is discussed considering both linear and turbo Successive Interference Cancellation ( SIC ) receivers. 

Q3. What future works have the authors mentioned in the paper "Transmission over multiple component carriers in lte-a uplink" ?

As a future work, the impact of bundling on the closed loop ( i. e. precoded ) transmission will be also evaluated, considering different precoding options ( e. g., per CC, per Antenna ). 

Q4. What is the effect of bundling on the closed loop?

Results show that, when a linear receiver is used in the BS, bundling per Antenna and bundling per CC can improve the spectral efficiency of the UE when combined with CC mixing and SD mixing, respectively, thus reducing the performance gap with no bundling. 

Q5. What is the effect of CC mixing on the UE performance?

When a turbo SIC receiver is used in the BS, CC mixing combined with bundling per Antenna can approximately achieve the performance of no bundling, whereas the SNR averaging over the antennas provided by SD mixing is shown to be detrimental. 

Q6. What is the intuitive way to cope with a multiple component carrier structure?

The most intuitive solution to cope with a multiple component carrier structure is to use a single DFT having dimension equal to the size of the transmit block over the whole used CC set. 

Q7. What is the way to improve the performance of the MIMO system?

The authors consider a 2x2 open loop MIMO system, as an expected candidate scheme for LTE-A uplink, and an effective transmission bandwidth of 10 MHz achieved by transmission over different numbers of CCs. 

Q8. What is the channel knowledge for the BS receiver?

Perfect channel knowledge is assumed at theBS receiver, for which the authors consider the following 2 options:1) Linear receiver: it is based on the traditional Minimum Mean Square Error (MMSE) equalization [4]. 

Q9. What is the purpose of the link adaptation process?

The issues related to the link adaptation process per CC will be discussed in Section V. For each CC, non-contiguous allocation of the Resource Blocks (RBs) on which the user data are scheduled has been approved with the aim of enhancing the scheduling flexibility. 

Q10. What is the effect of clustered allocation of RBs over the same CC?

Note that the clustered allocation of RBs over the same CC is highly detrimental for N=1 (allocation over 5 clusters performs even worse that 2CCs with 2 clusters), while this effect is considerably reduced when the number of CCs increases. 

Q11. What is the feedback overhead required for supporting the aforementioned solutions?

The feedback overhead required for supporting the aforementioned solutions is described in Table 1, assuming 10 MCSs’ options (therefore requiring 4 bits of feedback for indexing plus 1 bit for A/N message), and a spatial multiplexing system with 2 transmit antennas.