LTE downlink Physical layer transport mechanism (9)-centralized and distributed resource mapping

In LTE systems, the dynamic scheduling of RB resources is implemented on the ENB side, where the "RB resource" is actually referred to as virtualRB rather than physical RB(physical RB). VRB is the MAC layer used in the scheduling, is a logical concept, and PRB is the physical layer in the actual mapping of RE resources need to use, belonging to the actual physical sense of the concept . There are different mapping relationships between VRB and PRB: The simplest mapping relationship is that the location of VRB is the same as the location of PRB, which corresponds to one by one, and the other is that VRB and PRB are not one by one counterparts, but can rely on a particular mapping relationship. The location of PRB can be explicitly introduced through VRB. The first kind of simple relationship is the centralized resource allocation, and the mapping of the latter complex point is distributed resource allocation.

ENB side in the scheduling of resources, can be based on different scenarios using different allocation methods. For example, when using DCI2/DCI2A, the use of centralized resource allocation, while the use of dci1a, you can use both centralized resource allocation, or the use of distributed resource allocation. These two distributions are described in detail below.

The downstream VRB has two resource mappings: the VRB Resource mapping method of the centralized (localized type) and the VRB Resource mapping for distributed (distributed type). regardless of the type of VRB mapping, the VRB refers to "RB pair" (a pair of vrbs). We know that a sub-frame consists of 2 slots, a RB is composed of 12 sub-carriers in the frequency domain and a time slot on the time domain (refer to "LTE Physical Transmission Resources (3)-time-frequency resources" for the structure of RB), so for 12 sub-carriers on the frequency domain, a sub-frame corresponds to 2 RB, These two RB are called a "RB pair". When considering the VRB resource mapping, the 2 time slots in the same sub-frame RB is put together to consider, when using N_VRB to represent the RB number, refers to a VRB pair, rather than a specific time slot VRB. In other words, when describing the downstream resource mapping method, although the text is the location of VRB and PRB, in fact, it is the location of VRB to the prb pair of relations.

1. Centralized resource mapping (localized type)

In the centralized resource mapping mode, the VRB pair and the prb pair are one by one corresponding, that is, the location of VRB is the location of PRB, RB resource block ordinal N_PRB=N_VRB, the range is 0~ (n_dl_rb-1). is the location relationship of PRB and VRB when allocating VRB resources centrally, at 5MHz bandwidth.

2. Distributed resource Mapping (distributed type)

In the Distributed resource Mapping mode, theVRB pair and the prb pair are not one by one corresponding, the sequential VRB sequence number will be mapped to the discontinuous PRB sequence number, and the 2 slots within a sub-frame also have different mappings (as shown in 1), in this way to achieve " Distributed "Resource allocation. Whether to map a continuous VRB pair to a discontinuous prb pair , or to separate each prb pair, A prb pair of two RB resource transmission with a certain frequency interval (can be considered based on the time slot (slot) frequency hopping), the purpose is to achieve the frequency of the diversity effect.

Figure 1

Figure 1 above is a downlink bandwidth of 5MHz distributed resource allocation process, this diagram is only to help understand the distributed process, does not mean that the actual PRB mapping location is consistent with the picture. This diagram is simple and intuitive, and can describe the process very well, from the point of view of the Protocol to the specific mapping relationship, or how to get the actual mapping of prb pairs from VRB.

3. Get the Gap parameter

In the distributed distribution process, the PRB in 2 slots is mapped to different locations on the frequency domain, enabling time-slot-based frequency hopping. The Gap parameter in Figure 1 above is used to represent the frequency offset between a RB pair , measured in the number of RB units. Different bandwidth, this gap value is different, once the bandwidth is determined, the gap value is determined.

As can be seen from table 6.2.3.2-1 below, when the downstream bandwidth is 1.4MHz (n_dl_rb=6), 3MHz (n_dl_rb=15), 5MHz (n_dl_rb=25), only one gap value can be selected, the corresponding frequency offset value n_gap1 are: 1.4MHz n_gap1= 3, 3MHz n_ gap1=8, 5MHz n_ gap1=. The gap=12 in Figure 1 above conforms to the protocol for 5MHz bandwidth (n_dl_rb=25).

For the larger bandwidth of 10MHz (N_DL_RB=50), 15MHz (n_dl_rb=75), 20MHz (n_dl_rb=100), the protocol specifies two different gap values, whichever is selected, which is notified to the UE by the DCI. For distributed VRB allocations, such as the DCI1A format, there is a 1bit field (see yellow highlighting), which is used to indicate whether the current gap is using GAP1 or GAP2.

4. VRB range available for frequency interleaving (interleaving)

Not all VRB can be used for interleaving, and if N_VRB is used to represent the frequency position of the VRB, the VRB range that can be used for interleaving is 0~ (n_dl_VRB -1). Only the VRB in this range can be used for the interleaving of RB pairs to distribute the distributed VRB resources.

As shown, if the scope of the N_VRB is 0~10, then ENB for distributed resource scheduling, the RB ID number can only be assigned to 0~10 11 VRB. Note here the parameter n_dl_VRB is not the downlink bandwidth of RB number N_dl_RB, do not confuse. The parameter n_dl_VRB is calculated as follows:

(A) If GAP1 is used, then n_dl_vrb = N_dl_vrb_gap1 = 2xMin (n_gap1, n_dl_rb - n_gap1) For example, the current is 5MHz bandwidth,n_gap1=12,n_dl_rb= 25, then n_dl_vrb = 2xMin (12,25-12) = 2x12 = 24. If it is 10MHz bandwidth,n_gap1=27,n_dl_rb= 50, then n_dl_vrb = 2xMin (27,50-27) = 2x23 = 46. Therefore, for 10MHz bandwidth, using frequency interval GAP1 only 0~45 These 46 VRB pairs can be RB frequency interleaving, in order to carry out distributed RB distribution.

(b) if GAP2 is used, n_dl_vrb = N_DL_VRB_GAP2 = Floor (n_dl_rB/(2x n_gap2) /c4>) x2x n_gap2. For example, the current is 10MHz bandwidth,n_gap2=9,n_dl_rb= 50, then n_dl_vrb = Floor (50/18) x18 = 2x18 = 36.

5. Get the interleaving matrix (interleaving matrix)

In the process of frequency interleaving, the concept of interleaving matrix (interleavingmatrix) needs to be used, and the interleaving matrix is a matrix with a fixed number of columns equal to 4 . by interleaving The matrix it is easy to know where the VRB maps to where the PRB is.

The agreement stipulates that a continuous _n_dl_vrb VRB can form a interleaving unit (interleaving uint), each interleaving unit can be represented by a interleaving matrix with a number of elements of _N_DL_VRB . The parameter _N_DL_VRB can be obtained in the following way:

(A) If GAP1 is used, then _n_dl_vrb = n_dl_VRB. As can be seen from this formula, all VRB resources (i.e. n_dl_VRB values) that can be used for distributed scheduling can only form a interleaving unit, forming a interleaving matrix. For example, the current 5MHz bandwidth,n_dl_vrb=24, then a VRB interleaving unit consists of 24 consecutive VRB, that is, the interleaving matrix has 24 elements.

(B) If GAP2 is used, then _n_dl_vrb = 2x n_gap2. For example, the current is 10MHz bandwidth,n_gap2=9, then a VRB interleaving unit consists of 18 consecutive VRB, that is, the interleaving matrix has 18 elements. Because of the n_dl_Vrb=36at this time,that is , there can be 36 VRB for distributed resource allocation, so in this case, all the VRB available for distributed resource allocation can be composed of 2 interleaving units , generating 2 interleaving matrices.

As shown, the interleaving matrix for each interleaving unit is a 4-column,n_row -line matrix,n_row = ceil (_n_dl_vrb /(4xP)) XP , the value of P is shown in the table below. If the current is 5MHz bandwidth, p=2, with GAP1,_n_dl_vrb = N_DL_VRB = 24, then n_row = ceil (24 /(4x2)) x2 = 6.

VRB in the interleaving matrix , follow the rules for write-by-row and read-by-column . According to this rule, there is a case where the number of VRB that make up a interleaving unit is not sufficient to fill a interleaving matrix, so the Protocol stipulates that the last (null /2) row of columns 2nd and 4th needs to insert a null null value . The inserted null value is ignored when it is read out. Wherein,nnull = 4x n_roW-_n_dl_vrb. If the current is 5MHz bandwidth,_n_dl_vrb = N_DL_VRB = $,n_row = 6, then nnull = 4x6-24 = 0, which means that there is no need to insert null values at the end of columns 2nd and 4th, and all VRB exactly constitute a interleaving matrix, as shown in the schematic.

6.VRB mapping to PRB

With the interleaving matrix, you can map VRB to PRB based on the following 3 steps. These 3 steps are:

(1) from the interleaving matrix in order to read the elements in sequence, encountered Nnull Skip, remove a non-nnull element, so that is even the time slot prb reference position;

(2) The reference position of the dual-number time-slot is offset by (_n_dl_vrb /2), and the reference position of the odd time-slot prb is obtained;

(3) If the RB position obtained from the above two steps is greater than or equal to (_n_dl_vrb /2), then once again the offset is (Ngap- _ N_DL_VRB /2), resulting in the final PRB mapping position for even time slots and odd time slots.

After the above 3 steps, you can know vrb=3, 4, 5 three VRB, after distributed mapping, the location of PRB is (even time slot, odd time slot) = (18,6), (1,13), (7,19).

The above VRB to the PRB mapping process, can also use mathematical formula for strict derivation, the derivation process is as follows.

(1) The reference position of the even time slot is obtained according to the interleaving

The reference position _n_prb (ns) of the PRB corresponding to even-gap NS is calculated by the following formula:

(2) The position of the dual-number time slot (_N_DL_VRB /2) is offset to get the reference position of the odd time slot

The position _N_PRB (NS) of the odd timeslot NS is calculated from the following formula:

(3) The position of greater than or equal to (_N_DL_VRB /2) obtained in the previous two steps is shifted again (Ngap- _N_DL_VRB /2) to get the final prb position.

For all timeslot NS, including even and odd time slots, offset according to the following formula:

Based on the interleaving matrix, the mapping position of PRB has been obtained, and in order to verify with each other, the use process and calculation results of the mathematical formula are illustrated with 5MHz bandwidth.

(1) According to the relevant description of the previous article process, you can get n_gap1 =,_n_dl_vrb = ,n_row = 6,nnull = 0. Assuming that ENB assigns a vrb=3, 4, and 5 of three consecutive VRB to the UE, the process for calculating PRB locations is as follows:

(a) virtual resource to n_vrb=3, which occupies an even time slot and an odd time slot. As can be seen from the above formula, the PRB position of even time slots is only used to recursively calculate the number of time slots of the PRB position, in the actual calculation does not need to know the specific timeslot, the use of even time slots 0 and the use of time slots 2, 4, 6 and so on, the computed PRB mapping results are the same, So here can be time slot 0 for even timeslot, time slot 1 for odd timeslot. So:

For even-time slots (ns=0),_n_prb (0) = 6x ((3 MoD) mod 4) + Floor ((3 MoD)/4) + 24xfloor (3/24) = 18. Because the conditions of > (24/2) are met, the physical resource location n_prb (0) = + 12-24/2 = .

For odd time slots (Ns=1),_n_prb (1) = (_n_prb (0) + 24/2) mod + 24xfloor (3/24) = 6. Because the conditions of 6 < (24/2) are met, the physical resource location N_PRB (1) = 6, which does not need to be offset again.

(B) virtual resources to n_vrb=4:

For even-time slots (ns=0),_n_prb (0) = 6x ((4 mod) mod 4) + Floor ((4 MoD)/4) + 24xfloor (4/24) = 1. Because the conditions of 1 < (24/2) are met, the physical resource location n_prb (0) = 1.

For odd time slots (Ns=1),_n_prb (1) = (_n_prb (0) + 24/2) mod + 24xfloor (4/24) = 13. Because the conditions of > (24/2) are met, the physical resource location N_PRB (1) = + 12-24/2 = .

(C) virtual resources to n_vrb=5:

For even-time slots (ns=0),_n_prb (0) = 6x ((5 mod) mod 4) + Floor ((5 MoD)/4) + 24xfloor (5/24) = 7. Because the conditions of 7 < (24/2) are met, the physical resource location n_prb (0) = 7.

For odd time slots (Ns=1),_n_prb (1) = (_n_prb (0) + 24/2) mod + 24xfloor (5/24) = 19. Because the conditions of > (24/2) are met, the physical resource location N_PRB (1) = + 12-24/2 = .

ENB The allocation of resources, is the choice of centralized or distributed, is determined by the ENB side and through the DCI notification to the UE side . Different equipment manufacturers, the algorithm is different, the results of scheduling will vary. For the DCI1A format, where the field "localized/distributed VRB Assignment Flag" is used to denote the current VRB distribution (in Type2 distribution), Please refer to the post "LTE downlink Physical layer transport mechanism (5)-DCI format selection and dci1a". DC2, DC2A format VRB can only select centralized allocation (in TYPE0 or Type1 allocation).

Reference documents:

(1) 3GPP TS 36.211 V9.1.0 (2010-03) physical Channels and modulation

(2) 3GPP TS 36.212 V9.4.0 (2011-09) multiplexing and channel coding

LTE downlink Physical layer transport mechanism (9)-centralized and distributed resource mapping