Wednesday, March 30, 2011

Quick Recap of MIMO in LTE and LTE-Advanced

I had earlier put up some MIMO presentations that were too technical heavy so this one is less heavy and more figures.

The following is from NTT Docomo Technical journal (with my edits):

MIMO: A signal transmission technology that uses multiple antennas at both the transmitter and receiver to perform spatial multiplexing and improve communication quality and spectral efficiency.

Spectral efficiency: The number of data bits that can be transmitted per unit time and unit frequency band.

In this blog we will first look at MIMO in LTE (Release 8/9) and then in LTE-Advanced (Release-10)

MIMO IN LTE

Downlink MIMO Technology

Single-User MIMO (SU-MIMO) was used for the downlink for LTE Rel. 8 to increase the peak data rate. The target data rates of over 100 Mbit/s were achieved by using a 20 MHz transmission bandwidth, 2 × 2 MIMO, and 64 Quadrature Amplitude Modulation (64QAM), and peak data rates of over 300 Mbit/s can be achieved using 4×4 SU-MIMO. The multi-antenna technology used for the downlink in LTE Rel. 8 is classified into the following three types.

1) Closed-loop SU-MIMO and Transmit Diversity: For closed-loop SU-MIMO transmission on the downlink, precoding is applied to the data carried on the Physical Downlink Shared Channel (PDSCH) in order to increase the received Signal to Interference plus Noise power Ratio (SINR). This is done by setting different transmit antenna weights for each transmission layer (stream) using channel information fed back from the UE. The ideal transmit antenna weights for precoding are generated from eigenvector(s) of the covariance matrix of the channel matrix, H, given by HHH, where H denotes the Hermitian transpose.

However, methods which directly feed back estimated channel state information or precoding weights without quantization are not practical in terms of the required control signaling overhead. Thus, LTE Rel. 8 uses codebook-based precoding, in which the best precoding weights among a set of predetermined precoding matrix candidates (a codebook) is selected to maximize the total throughput on all layers after precoding, and the index of this matrix (the Precoding Matrix Indicator (PMI)) is fed back to the base station (eNode B) (Figure 1).


LTE Rel. 8 adopts frequency-selective precoding, in which precoding weights are selected independently for each sub-band of bandwidth from 360 kHz to 1.44 MHz, as well as wideband precoding, with single precoding weights that are applied to the whole transmission band. The channel estimation used for demodulation and selection of the precoding weight matrix on the UE is done using a cell specific Reference Signal (RS) transmitted from each antenna. Accordingly, the specifications require the eNode B to notify the UE of the precoding weight information used for PDSCH transmission through the Physical Downlink Control Channel (PDCCH), and the UE to use this information for demodulation.

LTE Rel. 8 also adopts rank adaptation, which adaptively controls the number of transmission layers (the rank) according to channel conditions, such as the received SINR and fading correlation between antennas (Figure 2). Each UE feeds back a Channel Quality Indicator (CQI), a Rank Indicator (RI) specifying the optimal rank, and the PMI described earlier, and the eNode B adaptively controls the number of layers transmitted to each UE based on this information.

2) Open-loop SU-MIMO and Transmit Diversity: Precoding with closed-loop control is effective in low mobility environments, but control delay results in less accurate channel tracking ability in high mobility environments. The use of open-loop MIMO transmission for the PDSCH, without requiring feedback of channel information, is effective in such cases. Rank adaptation is used, as in the case of closed-loop MIMO, but rank-one transmission corresponds to open-loop transmit diversity. Specifically, Space-Frequency Block Code (SFBC) is used with two transmit antennas, and a combination of SFBC and Frequency Switched Transmit Diversity (FSTD) (hereinafter referred to as “SFBC+FSTD”) is used with four transmit antennas. This is because, compared to other transmit diversity schemes such as Cyclic Delay Diversity (CDD), SFBC and SFBC+FSTD achieve higher diversity gain, irrespective of fading correlation between antennas, and achieve the lowest required received SINR. On the other hand, for PDSCH transmission with rank of two or higher, fixed precoding is used regardless of channel variations. In this case, cyclic shift is performed before applying the precoding weights, which effectively switches precoding weights in the frequency domain, thereby averaging the received SINR is over layers.

3) Adaptive Beamforming: Adaptive beamforming uses antenna elements with a narrow antenna spacing of about half the carrier wavelength and it has been studied for use with base stations with the antennas mounted in a high location. In this case beamforming is performed by exploiting the UE Direction of Arrival (DoA) or the channel covariance matrix estimated from the uplink, and the resulting transmit weights are not selected from a codebook. In LTE Rel. 8, a UE-specific RS is defined for channel estimation in order to support adaptive beamforming. Unlike the cell-specific RS, the UE specific RS is weighted with the same weights as the data signals sent to each UE, and hence there is no need to notify the UE of the precoding weights applied at the eNode B for demodulation at the UE. However, its effectiveness is limited in LTE Rel. 8 because only one layer per cell is supported, and it is an optional UE feature for Frequency Division Duplex (FDD).

Uplink MIMO Technology

On the uplink in LTE Rel. 8, only one-layer transmission was adopted in order to simplify the transmitter circuit configuration and reduce power consumption on the UE. This was done because the LTE Rel. 8 target peak data rate of 50 Mbit/s or more could be achieved by using a 20 MHz transmission bandwidth and 64QAM and without using SU-MIMO. However, Multi-User MIMO (MU-MIMO) can be used to increase system capacity on the LTE Rel. 8 uplink, using multiple receiver antennas on the eNode B. Specifically, the specification requires orthogonalization of the demodulation RSs from multiple UEs by assigning different cyclic shifts of a Constant Amplitude Zero Auto-Correlation (CAZAC) sequence to the demodulation RSs, so that user signals can be reliably separated at the eNode B. Demodulation RSs are used for channel estimation for the user-signal separation process.


MIMO TECHNOLOGY IN LTE-ADVANCED

Downlink 8-Layer SU-MIMO Technology

The target peak spectral efficiency in LTE-Advanced is 30 bit/s/Hz. To achieve this, high-order SU-MIMO with more antennas is necessary. Accordingly, it was agreed to extend the number of layers of SU-MIMO transmission in the LTE-Advanced downlink to a maximum of 8 layers. The number of transmission layers is selected by rank adaptation. The most significant issue with the radio interface in supporting up to 8 layers is the RS structure used for CQI measurements and PDSCH demodulation.

1) Channel State Information (CSI)-RS: For CQI measurements with up-to-8 antennas, new CSI-RSs are specified in addition to cell-specific RS defined in LTE Rel. 8 for up-to-four antennas. However, in order to maintain backward compatibility with LTE Rel. 8 in LTE-Advanced, LTE Rel. 8 UE must be supported in the same band as in that for LTE-Advanced. Therefore, in LTE Advanced, interference to the PDSCH of LTE Rel. 8 UE caused by supporting CSI-RS must be minimized. To achieve this, the CSI-RS are multiplexed over a longer period compared to the cell-specific RS, once every several subframes (Figure 3). This is because the channel estimation accuracy for CQI measurement is low compared to that for demodulation, and the required accuracy can be obtained as long as the CSIRS is sent about once per feedback cycle. A further reason for this is that LTE-Advanced, which offers higher data-rate services, will be developed to complement LTE Rel. 8, and is expected to be adopted mainly in low-mobility environments.


2) UE-specific RS: To allow demodulation of eight-layer SU-MIMO, the UE-specific RS were extended for SU-MIMO transmission, using a hybrid of Code Division Multiplexing (CDM) and Frequency Division Multiplexing (FDM) (Figure 4). The UE-specific RS pattern for each rank (number of layers) is shown in Figure 5. The configuration of the UE-specific RS in LTE-Advanced has also been optimized differently from those of LTE Rel.8, extending it for SU-MIMO as well as adaptive beamforming, such as by applying twodimensional time-frequency orthogonal CDM to the multiplexing between transmission layers.


Downlink MU-MIMO Technology

In addition to the peak data rate, the system capacity and cell-edge user throughput must also be increased in LTE-Advanced compared to LTE Rel. 8. MU-MIMO is an important technology for satisfying these requirements. With MU-MIMO and CoMP transmission (described earlier), various sophisticated signal processing techniques are applied at the eNode B to reduce the interference between transmission layers, including adaptive beam transmission (zero-forcing, block diagonalization, etc.), adaptive transmission power control and simultaneous multi-cell transmission. When these sophisticated transmission techniques are applied, the eNode B multiplexes the UE-specific RS described above with the PDSCH, allowing the UE to demodulate the PDSCH without using information about transmission technology applied by the eNode B. This increases flexibility in applying sophisticated transmission techniques on the downlink. On the other hand, PMI/CQI/RI feedback extensions are needed to apply these sophisticated transmission techniques, and this is currently being discussed actively at the 3GPP.

Uplink SU-MIMO Technology

To reduce the difference in peak data rates achievable on the uplink and downlink for LTE Rel. 8, a high target peak spectral efficiency of 15 bit/s/Hz was specified for the LTE-Advanced uplink. To achieve this, support for SU-MIMO with up to four transmission antennas was agreed upon. In particular, the two-transmission-antenna SU-MIMO function is required to satisfy the peak spectral efficiency requirements of IMT-Advanced.

For the Physical Uplink Shared Channel (PUSCH), it was agreed to apply SU-MIMO with closed-loop control using multiple antennas on the UE, as well as codebook-based precoding and rank adaptation, as used on the downlink. The eNode B selects the precoding weight from a codebook to maximize achievable performance (e.g., received SINR or user throughput after precoding) based on the sounding RS, which is used for measuring the quality of the channel transmitted by the UE. The eNode B notifies the UE of the selected precoding weight together with the resource allocation information used by the PDCCH. The precoding for rank one contributes to antenna gain, which is effective in increasing cell edge user throughput. However, considering control-information overhead and increases in Peak-to-Average Power Ratio (PAPR), frequency-selective precoding is not very effective in increasing system throughput, so only wideband precoding has been adopted.

Also, for rank two or higher, when four transmission antennas are used, the codebook has been designed not to increase the PAPR. The demodulation RS, which is used for channel estimation, is weighted with the same precoding weight as is used for the user data signal transmission. Basically, orthogonalization is achieved by applying a different cyclic shift to each layer, but orthogonalizing the code region using block spread together with this method is adopted.


Uplink Transmit Diversity Technology

Closed-loop transmit diversity is applied to PUSCH as described above for SU-MIMO. Application of transmit diversity to the Physical Uplink Control Channel (PUCCH) is also being studied. For sending retransmission request Acknowledgment (ACK) and Negative ACK (NAK) signals as well as scheduling request signals, application of Spatial Orthogonal-Resource Transmit Diversity (SORTD) using differing resource blocks per antenna or an orthogonalizing code sequence (cyclic shift, block spread sequence) has been agreed upon (Figure 6). However, with LTE-Advanced, the cell design must be done so that LTE Rel. 8 UE get the required quality at cell-edges, so applying transmit diversity to the control channels cannot contribute to increasing the coverage area, but only to reducing the transmission power required.

Monday, March 28, 2011

Friday, March 25, 2011

3GPP – DVB Workshop for Next generation Mobile TV standards

TSG RAN and TSG CT hosted a joint workshop with DVB project on commonalities between DVB-NGH and eMBMS

The workshop was opened by the RAN Chairman Mr. Takehiro Nakamura on Wednesday 16th March 11:07. This is the joint session between TSG RAN, TSG CT and DVB project expert. TSG CT Chairman Mr. Hannu Hietalahti reminded that the workshop can't make any formal decisions that would be binding on either 3GPP side or DVB project side. Any agreement needs to be confirmed in DVB project and 3GPP separately. From 3GPP side this needs to be done by 3GPP TSG RAN and 3GPP TSG CT meetings during this week. The goal of the workshop is to find a common agreement how to proceed the future work on DVB-NGH and eMBMS convergence and decide the best way forward. The joint session is expected to make recommendations to TSG SA #51 based on the service requirements for DVB-NGH and the commonalities with eMBMS that can be identified. TSG SA #51 will decide the best way forward on 3GPP side.

The MBMS presentation was embedded in this post. The DVB presentation is embedded below:



The minutes of the meeting are available here: http://3gpp.org/ftp/tsg_sa/TSG_SA/TSGS_51/Docs/SP-110185.zip

All the documents from this workshop are available here: http://www.3gpp.org/ftp/workshop/2011-03-16_RAN-CT-DVB/

It was agreed that for any 3GPP work the normal 3GPP working procedures should be used. The supporting 3GPP member companies were requested to initiate Study items in the appropriate 3GPP working groups with the aim of sending them for approval during the next Plenary cycle.

It was noted that 3GPP Rel-11 stage 1 is going to be frozen in September 2011. It was seen 3GPP DVB-NGH can be a part of Rel-11 if there are interest in 3GPP community. The interesting companies are expected to contribute according to 3GPP working procedures.

Interesting M2M Video by ETSI

Machine-to Machine Communications - David Boswarthick (15/02/2011) from ETSI – World Class Standards on Vimeo.

ETSI M2M: Building the Internet of Things

Presented by: David Boswarthick, ETSI Technical Expert

Live Presentation during MWC 2011: ETSI stand, Monday, 15 February 2011

_ _ _ _ _ _ _

About the presenter:

David Boswarthick, Technical Officer, ETSI

David has been extensively involved for over 10 years in the standardization activities of mobile, fixed and convergent networks in both the European Telecommunications Standards Institute (ETSI) and the 3rd Generation Partnership Project (3GPP). He is currently involved in the M2M standards group which is defining an end to end architecture and requirements for multiple M2M applications including Smart Metering, healthcare and enhanced home living. David holds a Bachelor's Honours Degree in Telecommunications from the University of Plymouth, and a Master's Degree in Networks and Distributed systems from the University of Nice and Sophia Antipolis, France.

Tuesday, March 22, 2011

3GPP Official 'MBMS support in E-UTRAN' - Mar 2011

Last month I blogged about the MBMS feature in Rel-9. The 3GPP official presentation on MBMS is now available. Embedded below:

Presentation can be downloaded from Slideshare.

This presentation was a part of Joint one hour session of 3GPP RAN and 3GPP CT on March 16th 2011, 11.00 am – 12.00 p.m. More on this coming soon.

Monday, March 21, 2011

A quick primer on Coordinated Multi-point (CoMP) Technology

From NTT Docomo Technical Journal:

CoMP is a technology which sends and receives signals from multiple sectors or cells to a given UE. By coordinating transmission among multiple cells, interference from other cells can be reduced and the power of the desired signal can be increased.

Coordinated Multi-point Transmission/Reception:

The implementation of intracell/inter-cell orthogonalization on the uplink and downlink in LTE Rel. 8 contributed to meeting the requirements of capacity and cell-edge user throughput. On the downlink, simultaneously connected UE are orthogonalized in the frequency domain. On the other hand, they are orthogonalized on the uplink, in the frequency domain as well as the code domain, using cyclic shift and block spreading. It is possible to apply fractional frequency reuse (A control method which assigns different frequency ranges for cell-edge UE) to control interference between cells semi-statically, but this is done based on randomization in LTE Rel. 8. Because of this, we are planning to study CoMP technology, which performs signal processing for coordinated transmission and reception by multiple cells to one or more UE, as a technology for Rel. 11 and later in order to extend the intracell/ inter-cell orthogonalization in LTE Rel. 8 to operate between cells.


Independent eNode B and Remote Base Station Configurations:

There are two ways to implement CoMP technology: autonomous distributed control based on an independent eNode B configuration, or centralized control based on Remote Radio Equipment (RRE) (Figure 7). With an independent eNode B configuration, signaling over wired transmission paths is used between eNode B to coordinate among cells. Signaling over wired transmission paths can be done with a regular cell configuration, but signaling delay and overhead become issues, and ways to increase signaling speed or perform high-speed signaling via UE need study. With RRE configurations, multiple RREs are connected via an optical fiber carrying a baseband signal between cells and the central eNode B, which performs the baseband signal processing and control, so the radio resources between the cells can be controlled at the central eNode B. In other words, signaling delay and overhead between eNode B, which are issues in independent eNode B configurations, are small in this case, and control of high speed radio resources between cells is relatively easy. However, high capacity optical fiber is required, and as the number of RRE increases, the processing load on the central eNode B increases, so there are limits on how this can be applied. For these reasons, it is important to use both distributed control based on independent eNode B configurations and centralized control based on RRE configurations as appropriate, and both are being studied in preparation for LTE-Advanced.

Downlink Coordinated Multi-point Transmission:

Downlink coordinated multi-point transmission can be divided into two categories: Coordinated Scheduling/ Coordinated Beamforming (CS/CB), and joint processing (Figure 8). With CS/CB, a given subframe is transmitted from one cell to a given UE, as shown in Fig. 8 (a), and coordinated beamforming and scheduling is done between cells to reduce the interference caused to other cells. On the other hand, for joint processing, as shown in Fig. 8 (b-1) and (b-2), joint transmission by multiple cells to a given UE, in which they transmit at the same time using the same time and frequency radio resources, and dynamic cell selection, in which cells can be selected at any time in consideration of interference, are being studied. For joint transmission, two methods are being studied: non-coherent transmission, which uses soft-combining reception of the OFDM signal; and coherent transmission, which does precoding between cells and uses in-phase combining at the receiver.

Uplink Multi-cell Reception:

With uplink multi-cell reception, the signal from a UE is received by multiple cells and combined. In contrast to the downlink, the UE does not need to be aware of whether multi-cell reception is occurring, so it should have little impact on the radio interface specifications.

Friday, March 18, 2011

Roadmap to Operational Excellence for Next Generation Mobile Networks


This presentation is from:

FP7 SOCRATES Final Workshop on Self-Organisation in Mobile Networks February 22, 2011 - Karlsruhe, Germany

This and all other presentations from this workshop are available to download from here.

Wednesday, March 16, 2011

Direct Communication between devices in case of disasters

Yesterday, a discussion started after I read this article on RCR Wireless News:

As in every major disaster, communications networks quickly showed their inherent weakness in times of greatest need. Japan's NTT Communications reported outages affecting Internet voice data that relies on IP-VPN technology.

In a brief statement, the operator apologized for the "trouble and inconvenience," following the string of earthquakes and significant aftershocks that rattled nerves and buildings throughout much of Japan. Some communication services are no longer available, NTT said, and telephone service, particularly long-distance service, is showing strain as well.

Service disruptions have been reported by all three of the major mobile operators in Japan, according to BusinessWeek.

This prompted me to ask on Twitter about which technologies are available that can help the mobile network cope with these problem.

Here are few approaches:

I blogged earlier about Multihop Cellular Networks (MCN) and ODMA. These technologies have their own limitations and problems and I have not heard of anything more about them being standardised or adopted.

Another post was on Ad-Hoc Networks that can be formed in case of failures resulting in Mobile devices being able to communicate directly without the need for network or base stations. The slight problem is that this approach replies on WiFi being available which may not always be the case.

A colleague suggested that in Tetra, Direct Mode of operation is available that is intended for situations like these. A presentation is embedded below:




Steven Crowley on twitter suggested that 802.16m has already started working in this direction. I got a related presentation on that which is embedded below:




Finally, Kit Kilgour mentioned about DSAC (Domain Specific Access Control) whose intention is to discontinue the voice service in emergency (to avoid congestion) but continue the packet domain normally. I have not looked at DSAC on this blog but in LTE instead Service Specific Access Control (SSAC) is used since LTE is PS only. See the blog entry here.

Please feel free to add any more information on this topic in the comments.

Monday, March 14, 2011

LTE Physical Layer Measurements of RSRP and RSRQ

One of the things on my mind for long time was to find a bit more about RSRP and RSRQ.

The following is from Agilent Whitepaper:

The UE and the eNB are required to make physical layer measurements of the radio characteristics. The measurement definitions are specified in 3GPP TS 36.214. Measurements are reported to the higher layers and are used for a variety of purposes including intra- and inter-frequency handover, inter-radio access technology (inter-RAT) handover, timing measurements, and other purposes in support of RRM.

Reference signal receive power (RSRP):

RSRP is the most basic of the UE physical layer measurements and is the linear average (in watts) of the downlink reference signals (RS) across the channel bandwidth. Since the RS exist only for one symbol at a time, the measurement is made only on those resource elements (RE) that contain cell-specific RS. It is not mandated for the UE to measure every RS symbol on the relevant subcarriers. Instead, accuracy requirements have to be met. There are requirements for both absolute and relative RSRP. The absolute requirements range from ±6 to ±11 dB depending on the noise level and environmental conditions. Measuring the difference in RSRP between two cells on the same frequency (intra-frequency measurement) is a more accurate operation for which the requirements vary from ±2 to ±3 dB. The requirements widen again to ±6 dB when the cells are on different frequencies (inter-frequency measurement).

Knowledge of absolute RSRP provides the UE with essential information about the strength of cells from which path loss can be calculated and used in the algorithms for determining the optimum power settings for operating the network. Reference signal receive power is used both in idle and connected states. The relative RSRP is used as a parameter in multi-cell scenarios.

Reference signal receive quality (RSRQ):

Although RSRP is an important measure, on its own it gives no indication of signal quality. RSRQ provides this measure and is defined as the ratio of RSRP to the E-UTRA carrier received signal strength indicator (RSSI). The RSSI parameter represents the entire received power including the wanted power from the serving cell as well as all cochannel power and other sources of noise. Measuring RSRQ becomes particularly important near the cell edge when decisions need to be made, regardless of absolute RSRP, to perform a handover to the next cell. Reference signal receive quality is used only during connected states. Intra- and inter-frequency absolute RSRQ accuracy varies from ±2.5 to ±4 dB, which is similar to the interfrequency relative RSRQ accuracy of ±3 to ±4 dB.

The following is from R&S white paper:


The RSRP is comparable to the CPICH RSCP measurement in WCDMA. This measurement of the signal strength of an LTE cell helps to rank between the different cells as input for handover and cell reselection decisions. The RSRP is the average of the power of all resource elements which carry cell-specific reference signals over the entire bandwidth. It can therefore only be measured in the OFDM symbols carrying reference symbols.

The RSRQ measurement provides additional information when RSRP is not sufficient to make a reliable handover or cell reselection decision. RSRQ is the ratio between the RSRP and the Received Signal Strength Indicator (RSSI), and depending on the measurement bandwidth, means the number of resource blocks. RSSI is the total received wideband power including all interference and thermal noise. As RSRQ combines signal strength as well as interference level, this measurement value provides additional help for mobility decisions.

Assume that only reference signals are transmitted in a resource block, and that data and noise and interference are not considered. In this case RSRQ is equal to -3 dB. If reference signals and subcarriers carrying data are equally powered, the ratio corresponds to 1/12 or -10.79 dB. At this point it is now important to prove that the UE is capable of detecting and decoding the downlink signal under bad channel conditions, including a high noise floor and different propagation conditions that can be simulated by using different fading profiles.

I will be adding some conformance test logs at the 3G4G website for Measurement and Cell Selection/Re-selection that will give some more information about this.

In case you can provide a much simpler explanation or reference please feel free to add in the comment.