A couple of days ago I had a chat with one of my colleagues about overall antenna / system configuration for MIMO and Carrier Aggregation. After the chat, he asked me if I can recommend any books or technical papers that meets following criteria
- Minimized written description
- Easy enough so that any beginners can understand
I know these are the wishlist for anybody (including me) who tries to learn on any new technologies, but in reality it would be almost impossible to write a technical documents who perfectly meet these simple (?) criteria. 'Minimized written description' often omits a lot of details and omitting the details often need to sacrifice the accuracy. However, I think it is always worth trying to write this kind of document just to provide some big pictures for those who just got into the area even with sacrifice of some accuracy.
Since I don't recall any books and documents that I would recommend at the time, I decided to try writing my own version. I think I can easily meet the first criteria in this page, but no guarantee for the second criteria :)
Follwoings are the list of topics that I will cover in this page and I think (hope) this list will go on and on as I get more feedback from readers.
- What does it mean ? Notation
- How they are connected ?
- Connections in Test Equipment - eNB Simulator
- Carrier Aggregation and MIMO
- Example : 2CC CA : 2x2, 2x2
- Example : 2CC CA : 4x4, 4x4
- Example : 3CC CA : 4x4,4x4,2x2
- Example : 4CC CA : 4x4,4x4,4x4,4x4
- Example : 5CC CA : 4x4,4x4,4x4,4x4,4x4
- Carrier Aggregation Band Combination : Intra vs Inter band
- TB - Transport Block, CW - Code Word, Layer
- Reference
What does it mean ? Notation
The first thing you need to get familiar with in learning any new technology is to get used to some termnilogy. In terms of MIMO, you might have heard so often of 2x2 MIMO, 4x4 MIMO, 4x2 MIMO etc. I think everybody would knows that MIMO stands for Multi Input Multi Output. Then you might ask 'what does Input mean and what is Output mean ?'. Technically Input / Output mean that Input mean the MIMO processor Input and Output mean the MIMO processor Output. Then you may ask what is the MIMO processor.. so this kind of Q&S would goes on and on. So just for the simplicity (with the sacrifice of a little bit of accuracy), think of Input as a transmittor antenna and output as reciever antenna.
Then what does 2x2,4x2,4x4 mean .. those numbers indicates the number of Tx antenna and the number of Rx antenna as illustrated below. I don't think you would need any further explanation about this.
DL MIMO. The first number counts the eNB transmit antennas, and the second counts the UE receive antennas.
UL MIMO. The same notation, but now the UE transmits and the eNB receives.
The order of the two numbers is therefore fixed by the direction, not by the device. The first number is always the transmit side, and the second number is always the receive side. The notation says nothing about the number of layers. The number of layers cannot exceed the number of transmit antenna ports (36.211 clause 6.3.3.2), and the receiver needs at least as many antennas to separate them. So the smaller of the two numbers is the upper limit on the layers.
With this in mind, just go through some of the examples as shown below and see if they make sense to you
Example : 2x2 DL MIMO
The 2x2 DL case is the basic LTE MIMO configuration. The eNB transmits on two antennas and the UE receives on two, so the link can carry two layers. 36.306 Table 4.1-1 gives two layers to every UE category from Category 2 upward.
2x2 DL MIMO: 2 Tx antennas on the eNB and 2 Rx antennas on the UE.
Example : 4x2 DL MIMO
With 4x2, the eNB has four transmit antennas but the UE still has only two receive antennas. The UE can take at most two layers. The two extra eNB antennas add transmit diversity or precoding gain, not more layers.
4x2 DL MIMO: 4 Tx antennas on the eNB and 2 Rx antennas on the UE.
Example : 4x4 DL MIMO
4x4 needs four receive antennas in the UE, so it came with the higher UE categories. 36.306 Table 4.1-1 gives Category 5 up to 4 layers, and Categories 6, 7, 9 and 10 support 2 or 4 layers.
4x4 DL MIMO: 4 Tx antennas on the eNB and 4 Rx antennas on the UE. This configuration can carry up to 4 layers.
Example : 2x2 UL MIMO
The same notation applies to the uplink, but the roles swap: the UE transmits and the eNB receives. Uplink spatial multiplexing came in Release 10, and 36.211 clause 5.3.2A allows up to four layers on the PUSCH.
2x2 UL MIMO: 2 Tx antennas on the UE and 2 Rx antennas on the eNB. The colours swap with the DL images, because red marks the transmitter.
P x Q : P transmit antennas, Q receive antennas.DL : the eNB transmits; UL: the UE transmits.Layers : at most the smaller of P and Q.
How they are connected ?
Now I think (hope) you got the general understanding on the MIMO antenna configuration. And then you may ask 'how the signal from the Tx antenna travel to Rx antenna'. Unfortunately this is not a question that can easily be answered. but I will try.
In live(real) network, the signal coming out of the Tx(Transmitter) antenna spread through every directions and reaching the Rx(Reciever) antenna via various different path as shown on the left side. Depending on the communication environment, the path would get more complicated or simpler. However, you wouldn't see this kind of connection diagram often in MIMO technical documents. What you would see in most of technical documents would be something like the one on the right side. However, the path on the right side is a kind of mathematical representation and it is not a physical representation. Don't think too much of the words 'mathematical connection' too much. Just think that each of the arrows indicates the possible signal flow from Tx to Rx antenna (Each of the single arrow has much more complicated meaning than you may think.. but I will not get any further into this. Because of these complexity, it is very difficult to implement the MIMO signal path using direct cable connection).
Left: the physical paths in a real network, with reflections from buildings, trees and hills. Right: the logical paths that textbooks draw. Each arrow on the right stands for all the physical paths between one Tx antenna and one Rx antenna.
Each arrow on the right is one element of the channel matrix H. For 2x2 MIMO there are four elements, h11, h12, h21 and h22. Each is a complex gain that sums every physical path between one Tx antenna and one Rx antenna. The receiver estimates these elements from the reference signals of each antenna port, and then separates the layers with them.
MIMO works best when the elements of H differ from each other. If all four paths look the same, the receiver cannot tell the layers apart, and the link falls back to fewer layers. A rich scattering environment, like the one on the left, is therefore good for MIMO.
Arrows are not cables : each one is an element of the channel matrix H.P x Q MIMO has P x Q elements : four elements for 2x2.Different paths help : similar paths make the layers hard to separate.
Connections in Test Equipment - eNB Simulator
If you are just a MIMO user (e.g, mobile phone user), you wouldn't even care of how those Tx and Rx antenna are interacts. However, if you are involved in the industry of MIMO development and application, you would consider MIMO mostly in the context of simulated environment (e.g, Using eNB simulator and UE(DUT). Most of the MIMO test requires is done in conductive environment (i.e, in RF cable connection). However, as mentioned above it is very difficult (almost impossible) to implement the MIMO signal path by direct cable connection. Then how the equipment provide such a way that enables cable connection between eNB and UE(DUT) in MIMO setup ?
Taking 2x2 MIMO as a simple example, it can be illustrated as shown below (As far as I understand, most of UE test equipment in the market implement MIMO in similar way as explained here). As you see here, the MIMO path (mathematical MIMO path) is simulated inside the equipment and only the resulting output of the MIMO path are connected to the equipment RF port. You might have thought the RF port labeled as Out port on the equipment is not the Tx Antenna port in MIMO operation. Actually it functions as Rx Antenna port in MIMO diagram shown in previous section. So just connecting UE Rx port to the equipment Out port as shown below allows us to perform MIMO test with RF cable connection between DUT and the equipment. (Don't get disappointed even if this does not make clear sense to you right away. It would not be simple to understand. Just try to see this illustration several times with some interval (like several days or a couple of weeks), it will start making sense to you gradually).
2x2 DL MIMO in a test set. The MIMO path is simulated inside the equipment, and the RF ports act as the UE Rx antennas.
By the same logic, 4x2 and 4x4 can be implemented as shown below. I hope these make sense to you.
4x2 on the left and 4x4 on the right. The equipment has one output port per UE Rx antenna, not per eNB Tx antenna.
The number of cables therefore follows the UE, not the eNB. A 4x2 test needs two cables and a 4x4 test needs four. For conformance tests, 36.101 does not leave the simulated path to the vendor. Annex B.2.3 defines MIMO correlation matrices at high, medium and low levels, and the performance tests name the level they use.
The MIMO path is simulated in firmware : the RF cables carry the result.One port per UE Rx antenna : 2 cables for 4x2, 4 for 4x4.36.101 Annex B.2.3 : defines the correlation levels for conformance tests.
Carrier Aggregation and MIMO
At a very high level view, Carrier Aggregatioin(CA) and MIMO has the same goal. It is to transmit the multiple data stream in parallel (i.e, simultaneosly) to achieve very high data throughput. The difference between MIMO and Carrier Aggregation. Actually there is another way of simultaneous throughput of multiple data stream, called CoMP(Coordinated multi-point operation). However, CoMP is not in the scope of this page (Refer to CoMP page if you are interested in ). Following illustration would would give you a comparative pictures on MIMO, CA, CoMP.
Just focusing on the comparision between MIMO and CA, you can say 'MIMO is the transmission of the multiple data stream within one eNB(BTS) via a same frequency and CA is the transmission of the multiple data stream using multiple eNB(BTS) with each eNB having different frequencies. In most case, MIMO and CA are used at the same time (i.e, Multiple eNB are transmitting multiple streams and each of the eNB in the CA setup is using MIMO).
The images below draw one eNB box per component carrier to keep the picture simple. In LTE CA, however, the component carriers are cells of the same eNB (36.300 clause 5.5). Aggregating carriers from two different eNBs is a separate feature, dual connectivity.
From SISO to MIMO, and then two ways to go further: carrier aggregation on different frequencies, and CoMP, usually on the same frequency.
Here goes some examples of CA with MIMO. I think (hope) the illustration itself would be self-explanatory. I wouldn't put any additional comments for each example to minimize reading :). Following examples are the most common CA/MIMO configurations that you might have seen (or hear about).
Example : 2CC CA : 2x2, 2x2
Two component carriers with 2x2 MIMO on each give the UE four layers in total, two per carrier. Category 6 reaches its peak of 301504 bits per TTI in this configuration, with 20 MHz and 64QAM on each carrier.
2CC CA: 2x2 MIMO on the PCC and 2x2 on the SCC, which is 4 layers in total.
Examples of throughput estimation for this configuration are as follows :
Example : 2CC CA : 4x4, 4x4
With four receive antennas on each carrier, the same two carriers carry eight layers instead of four. The UE needs four receive antennas that work on both bands.
2CC CA: 4x4 MIMO on the PCC and on the SCC, which is 8 layers in total.
Examples of throughput estimation for this configuration are as follows :
Example : 3CC CA : 4x4,4x4,2x2
Mixed configurations are common, because not every band has four eNB antennas. Here the PCC and SCC1 run 4x4 and SCC2 runs 2x2, which gives 4 + 4 + 2 = 10 layers.
3CC CA: 4x4 on the PCC and SCC1, and 2x2 on SCC2, which is 10 layers in total.
NOTE : It would be obvious on how each of antenna are mapped between eNB and UE when the number of eNB antenna and the number of UE antenna are same as in PCC and SCC1. Then you may ask how the antenna would map for SCC2 where the number of eNB antenna is smaller than the number of UE Rx antenna ? In this case, the two antenna from SCC2 are mapped to the first two antenna of the 4 UE antenna.
Examples of throughput estimation for this configuration are as follows :
Example : 4CC CA : 4x4,4x4,4x4,4x4
Four carriers with 4x4 MIMO on each give 16 layers. The image header reads 5CC, but the drawing and the list below show four carriers: the PCC and SCC1 to SCC3.
4CC CA: 4x4 on each of 4 carriers, which is 16 layers. The header in the image says 5CC by mistake.
Examples of throughput estimation for this configuration are as follows :
Example : 5CC CA : 4x4,4x4,4x4,4x4,4x4
Five carriers was the Release 10 limit, set by maxServCell-r10 = 5 in 36.331. With 4x4 on every carrier, this example reaches 20 layers. Release 13 raised the limit to 32 serving cells with maxServCell-r13.
5CC CA: 4x4 on each of 5 carriers, which is 20 layers.
CA adds carriers, MIMO adds layers per carrier : the total is the sum over the carriers.All CCs belong to one eNB : carriers from two eNBs are dual connectivity.5 CCs in Release 10 : 32 serving cells from Release 13.
Carrier Aggregation Band Combination : Intra vs Inter band
Theoretically you can assign any frequency and any LTE Bandwidth for each CC(Component Carrier) for a CA(Carrier Aggregation) configuration. So if the number of CC in a CA gets larger the possible combination of carrier frequencies would go almost infinite. But 3GPP defines only a limited set of carrier combinations and the network operator would chose a specific combination from the 3GPP list(These list is defined in 3GPP TS 36.101). Even thought the list is too long for you to memorize, it would be helpful if you know of a couple of important pattern (categories) for carrier allocation.
In terms of carrier allocation pattern, you migh have heard often about 'intra band' or 'inter band'. Basically these terminology is used to describe the band allocation between two carriers. 'Inter' means 'between'. So 'Inter band' means 'between (different) bands' implying that all of the carriers belong to different band. 'Intra' means 'within'. So 'Intra band' means 'within a (same) band' implying that all of the carriers belong to a same band.
Intra band allocation has two different pattern called 'Intra band contiguous' and 'Intra band non-contiguous'. 'Contiguous' mean 'Sitting right next to each other' and you don't need any further explanation for 'Intra band non-contiguous'.
36.101 writes a CA configuration as the band numbers, each followed by a CA bandwidth class letter. The class says how many contiguous carriers the band holds and how many RB they add up to. So CA_1A-3A is inter band with one carrier in Band 1 and one in Band 3. CA_1C is intra band contiguous with two carriers in Band 1, and CA_3A-3A is intra band non-contiguous with two separate carriers in Band 3.
CA bandwidth class | Aggregated RB | Contiguous CC |
A | up to 100 | 1 |
B | 26 to 100 | 2 |
C | 101 to 200 | 2 |
D | 201 to 300 | 3 |
E | 301 to 400 | 4 |
F | 401 to 500 | 5 |
I | 701 to 800 | 8 |
These are the classes of 36.101 v20.0.0 Table 5.6A-1. A UE reports the combinations it supports in its capability, and the network can only configure one of them.
I think you would have pretty good understandings on carrier allocation pattern by now, but several examples illustrated as below would give you much clearer / intuitive understanding.
Example : 2CC CA - Inter band
Inter band: the PCC in Band A and the SCC in Band B. 36.101 writes this type like CA_1A-3A.
Example : 2CC CA - Intra band, Non-contiguous
Intra band non-contiguous: both carriers in Band A, with a gap between them. 36.101 writes this type like CA_3A-3A.
Example : 2CC CA - Intra band, contiguous
Intra band contiguous: both carriers in Band A, next to each other. 36.101 writes this type like CA_1C.
Example : 3CC CA - All carriers are in Inter band relationship.
3CC inter band: one carrier in each band. 36.101 writes this type like CA_1A-3A-7A.
Example : 3CC CA - PCC and SCC1 are in intra band continguous relationship. PCC/SCC1 and SCC2 are in inter band relashionship.
3CC mixed: a contiguous pair in Band A and one carrier in Band C. 36.101 writes this type like CA_1A-3C, with the class C band holding the pair.
Example : 5CC CA - (SCC1 and SCC2) and (SCC3 and SCC4) are in intra band continguous relationship. PCC and other bands are in inter band relashionship.
5CC mixed: the PCC in Band A, and contiguous pairs in Bands B and C. 36.101 writes this type like CA_1A-3C-7C.
Inter band : carriers in different bands.Intra band contiguous : one class letter B to F in one band.Intra band non-contiguous : the same band repeated, as in CA_3A-3A.
TB - Transport Block, CW - Code Word, Layer
The main reason I put this section is that some equipment vendor specifies their MIMO / CA capability in terms of Layers. For example, you may see the equipment description saying "Capable of supporting max 10 layers" etc.
Followings are examples showing the overall data flow in MIMO / CA indicating the relationship among TB, CW, Layer (Detailed procedure of each steps are very complicated.. just follow through the terminologies for now).
NOTE : I often hear people talk about 'Stream', but it has a little bit vague meaning and people from different technical background may use it in a little bit different meaning. In LTE, Stream is often used as the same meaning as Layers, but be careful when you are talking to others.
NOTE : All of the illustrations shown below are showing the transmitter side only. It is assumed that the reciever side has the same number of Rx antenna)
NOTE : However many antenna are used in an eNB(BTS), the maximum number of TB(Transport Block) and CW(Codeword) are always 2.
The limit of two applies per carrier. 36.211 v19.3.0 Table 6.3.3.2-1 maps at most two codewords to up to eight layers. With CA, each carrier has its own transport blocks, so a UE with two carriers can receive four transport blocks in one TTI, as the last example below shows.
Example : 2x2 MIMO
2x2 MIMO: 2 TBs, 2 CWs and 2 layers. Each codeword maps to one layer.
Example : 4x4 MIMO
4x4 MIMO: 2 TBs, 2 CWs and 4 layers. CW0 goes to layers 0 and 1, and CW1 to layers 2 and 3, as in 36.211 Table 6.3.3.2-1.
Example : 2CC CA - 4x4 + 4x4 (PCC 4x4, SCC 4x4)
2CC CA with 4x4 on each carrier: 4 TBs, 4 CWs and 8 layers in total.
A vendor figure such as "max 10 layers" is therefore a sum over the carriers. The 3CC example above, with 4x4, 4x4 and 2x2, is one way to reach 10 layers. Within one carrier, 36.306 Table 4.1-1 gives 8 layers to Category 8, which is the LTE maximum.
Up to 2 TBs and 2 CWs per carrier : 36.211 Table 6.3.3.2-1.Up to 8 layers per carrier : Category 8 in 36.306.Totals add up over the carriers : 2CC with 4x4 gives 4 TBs and 8 layers.
Reference
[1] 3GPP TS 36.211 v19.3.0 - clause 6.3.3, Layer mapping, and clause 5.3.2A, uplink layer mapping
[2] 3GPP TS 36.300 v19.2.0 - clause 5.5, Carrier Aggregation
[3] 3GPP TS 36.101 v20.0.0 - Table 5.6A-1, CA bandwidth classes, and Annex B.2.3, MIMO correlation matrices
[4] 3GPP TS 36.306 v19.3.0 - Table 4.1-1, Downlink physical layer parameter values set by ue-Category
[5] 3GPP TS 36.331 v19.3.0 - maxServCell-r10 and maxServCell-r13