From the SelectedWorks of Ekta Gujral Miss

Winter 2013

Robust PCI Planning for Long Term Evolution Technology Ekta Gujral, Miss

Available at: http://works.bepress.com/ekta_gujral/1/

Robust PCI Planning for Long Term Evolution Technology

Abstract Long Term Evolution is one part of the 4G technology which targets the radio network Evolution. LTE is specified from Release 8 and 3GPP release 13 is available now days. The goal of LTE is to speed up the network and increase the capacity. The features of LTE technology is high data rates (DL > 100Mbps and UL > 50 Mbps). Spectral efficiency (1.4 MHz, 3 MHz, 5 MHz, 10 MHz, 15MHz, 20 MHz) make LTE robust. It is simple technology because of less signaling. The eNodeB is auto configurable. Every eNodeB have its own PCI (Physical Cell ID). According to 3GPP guidelines there are 504 PCIs available for the network. This paper demonstrates the concept how we can assign a robust PCI to network. And also describe the effect on network if the wrong PCI is assigned to eNodeB. To check the robustness of our algorithm, we use 1484 km2 area having 1964 sites. Introduction The LTE radio interface is based on OFDM (Orthogonal Frequency Division Multiplex) and OFDMA (OFDM Access) in DL and SC-FDMA (Single Carrier Frequency Division Multiple Access) in UL. These techniques are well suited for flexible bandwidth operation. This enables operators to deploy LTE in different regions with different frequency bands and bandwidths available. So for this network, PCI planning is more importance because if Optimal PCI is not assigned to eNodeB the signals will overshoot and cause interference. First of all discuss the concept of PCI. The PCI is combination of two synchronization signals. 1. PSS ( primary synchronization signals, PCI- ID ) 2. SSS ( Secondary synchronization signals , PCI- group) PSS is Primary Synchronization Signal (PSS) is present in subframe 0 and 5 (OFDM symbol 6) and mapped on 72 subcarriers in the middle of the band. The PSS is 3 sequence number (0, 1 and 2) Secondary Synchronization Signal (SSS) is present in subframe 0 and 5 (OFDM symbol 5) and also mapped on 72 subcarriers in the middle of the band as shown in figure 1.The SSS is 168 sequence number (0 to 167).

Figure 1: Position of PSS and SSS in One frame (10ms) PCI is combination of PSS and SSS. The 168 Physical-Layer Cell-Identity groups with 3 PhysicalLayer Identities per group makes 168 × 3 = 504 Physical-Layer Cell Identities (PCI). Mathematically: PCI = PSS + 3*SSS

So for LTE network we have PCI 0 (PSS = 0, SSS = 0) to PCI 503 (PSS =2, SSS = 167). PCI Assigning Problem Formulation Let's say that we are going to deploy a LTE network in a city and that city needs 6000 cells. Each of the 6000 cells will have their own PCI, but since there is only 504 physical cell IDs, we will need to repeat the physical cell IDs. The key is that the two cells that share a PCI cannot be geographically close to each other otherwise they will interfere will each other. Very briefly; it is very expensive to use RF DTs at each cell location to determine cell overlap and interference. A better approach is predictable in advance and such algorithm is implemented that two cells having same PCIs should not close to each other. Once assigned, cells need not constantly reconfigure when the network is expanded thus the network remains stable as concerns IDs. (No reassignment influenced by a neighboring cell change) There is more that operators can do to manage RF (QAM) and physical IDs from a centralized perspective within the specifications, including client density and cell throughput via algorithms planned before a deployed network and layered on RF spectrum limitations There are two things that are needed to keep in mind that eNodeB have collision free and confusion free PCI. Collision free PCI means that two cells that are adjacent to each other do not have same PCI and Confusion free PCI means that the cell may not have neighbour with same PCI. Also Frequency shift plays very important role during the PCI assignment. The PCI itself gives the frequency shift through the formula given below: P0 = PCI mod 6 + k.6 P1 = (PCImod6) mod3 +k.6 Where k = 0 or 1, P0= 1st reference Signal position and P1 = 2nd reference signal position. So eNodeB with same frequency shift cause interference.

Figure 2: PCI Assignment Problem of Collision and Confusion

. Approach For the green field network, First consideration is that you need to clear the dense area. Most of the clashes of physical ids occurred in dense areas. Now question arises how we can find the dense area the logic for PCI assignment is divided into the following broad categories:• • • •

Identifying the Number of PCI’s to be used and the PCI Spacing. Identification of the first site to be selected for PCI Assignment. Allocation of PCI to the first batch (cluster) of sites. PCI Assignment for the remaining network using Min. Tier and Max. Distance Concept and maintaining a uniform PCI RE-USE pattern.

Identifying available PCI pool Out of the Total available pool of 504 PCI, some can be reserved for the future use. So some SSS are reserved and the network is planned using the remaining available SSS.

Figure 3: Color group for PCI pool The user may specify the number of Color Code Groups to remain reserved and unassigned to allow for future manual planning. As each Site Group contains 3 PCIs, the number of reserved PCIs will be 3 times the value entered. Groups are reserved from the bottom. PCI Spacing The sector spacing of 1, 4, and 8 can be used to plan the sites as per the design requirement.

Figure 4: Pci Spacing 4

Figure 5: Pci Spacing 8

Before applying any algorithm, there is need of one data set in which site to site distance is defined. With the help of this table we fix the distance issues during assignment.

Table1: NNI table This calculation is done for every site in the network. If there are N sites in the network then for all the N sites total (N-1) relationships will be made.

How to find the densest Site? The Dense area identification helps in clean PCI Assignment. For PCI assignment we select the first densest site from the network. Solution to the above question is explained in diagram. Draw circle of 2 to 3 KM around each and every site. Then count the number of sites coming in that circle. The site which has highest count is actually densest site. .

Figure 6: Site count for each Site within 3 KM radius The following is done for every site in the given network and a table matrix is created as follows: Max. Site Count.

Table 2: Site count for each Site for network The site with maximum site count is selected as the first site to be given PCI 0. Assigning PCIs to densest area: There are two possibilities that user want to use all SSS or want to reserve some of them. So for this paper we keep 20 SSS reserve. Now we left with 148 SSS. And 1 SSS is already used for densest site. Next step is to find the 147 nearest sites to densest site. Assign next SSS to these sites. Densest cluster have unique 148 SSS. The advantage of planning first dense cluster is that cluster has collision and confusion free PCIs.

Table for nearest 147 sites (Distance based)

Table 3: Assignment of PCI to 1st group of sites After allocation of PCI 0(sector 1) to the first site we assign the subsequent PCI 3,6,9… so on to the next set of sites on the basis of closest distance from site 1. The total available PCIs 0 to 444(total 148 for sector 1 if 60 reserved PCIs) are assigned. PCI Assignment to the next sites After the assignment of first 148 sites, we start assigning the PCI from 148th site in the table. Now the 149th site is taken as reference and we identify the next set of 148 sites to be assigned PCI on the basis of closest site (Min. distance from 148th site).The sites are selected in batches of 148 and the whole network is covered this way. List of sites with Assigned PCI As well as non-assigned site (PCI= -1).

Next target site (149th)

Table 4: Searching for 2nd unassigned group of sites

After assignment of PCI to first 148 sites and identification of the next set of 148 sites to be assigned the following logic is implemented for the 149th site (first site in the list of Un-assigned sites) which inTABLE-4 is site SKDVRRX. The table of assigned PCI is scanned and the PCI lying at Max. Distance is the chosen PCI. The assignment of PCI is updated in the Assigned PCI Table. So using maximum distance concept we assign PCIs to second cluster of 148 Sites. Now the challenge comes, because in network most of the PCIs are assigned more than 1 time. If for next group of sites we use the maximum distance concept then the confusion and collision chances are more. For further assignment of PCI to sites, the following logic is implemented. The assigned PCI table is as follows:-

PCI 490 RE-USED two times

Table 5: Assigned sites in network From this Table we filter out one entry per PCI. The entry with MAX (Min. distance per PCI) from the given site and satisfying the criteria (tier >3) is the allocated PCI. This logic is repeated for the entire network. In case there is no available PCI that satisfies the above criteria, the site will get PCI= -1 (Dummy value) and the tool can be re-run again for assignment to sites which are unassigned. Update the PCI to sector 2 and 3 according to the spacing chosen by user. Mathematically: PCI = ([PCI Spacing] *[PSS]) + ([SSS]*3) For example, for PCI spacing 8 and SSS = 1, PCI is calculated as: Sector 1: 8*0 + 3*1 = 3 Sector 2: 8*1 + 3*1 = 11 Sector 3: 8*2 + 3*1 = 19

The whole process is explained in flow chart also.

Result and Analysis To check the robustness of our algorithm we use SQL and VB.Net plateform. Using this algorithm we see thatv there is no any PCI which have tier below 3. Avgerage distance per PCI for our network is 21.8 Km. We check the stats for Avg Distance per PCI(only sector 1), No. of sites per Tier and No. of Sites per PCI.The graphs and tables for results are given below: •

Avg. Distance Per PCI

Table 6 shows the result for average diatnce per PCI(only first12 Pci shown in table). From the grapgh view we can say that for whole network Avg distance is almost linear. The Avgerge distance this network is 21.8 Km for Pci reuse.

Table 6: Avg dist per PCI •

Grapgh1 : Avg dist per PCI

No of sites Per PCI

Table 7 shows the result for No of sites per PCI(only first12 Pci shown in table). From the grapgh view we can say that for whole network Pci reuse is almost constant The Avgerge Reuse for this network is 13

Table 7: No of Sites per PCI

Grapgh2 : No. of Sites per PCI

•

No of sites Per Tier\

To Avoid collision and confussion we avoide to give same pci to site which have less than 3 sites at LOS. So from our algorithm the maximum number of sites have tiers between 6 to 48. The results are given below.

Table 8: No of Sites per tier

Grapgh3 : No. of Sites per tier

Conclusion Above results shows the robust asignment of PCI to network. IF PCI is not accuratelly assigned to network then interferance is hogh and data services will effected too much. Due to PCI clash in network UE start losing RRC connection which is result of high drop rate in network.Hence with this algorithm there is no confussion and cillison for two site with respest to PCI.