1X3 Frequency Reuse Technology Guideline (v2.0)
1X3 Frequency Reuse Technology Guideline (v2.0)
1X3 Frequency Reuse Technology Guideline (v2.0)
0)
Public Use
Document No.
Product version
Confidentiality level
24 pages in total
2005-01-18
Public Use
Table of Contents
1 General..........................................................................................................................................................3 1.1 Application background...................................................................................................................4 1.2 Basic concepts..................................................................................................................................4 1.3 The advantages and disadvantages of 1*3.......................................................................................5 1.3.1 Advantages............................................................................................................................5 1.3.2 Disadvantages.........................................................................................................................5 2 1*3 tight reuse technology..........................................................................................................................6 2.1 Layout of sites .................................................................................................................................6 2.2 1*3 improve the capacity of network...............................................................................................7 2.3 1*3 tight reuse pattern......................................................................................................................8 2.3.1 The basic concepts of frequency hopping.............................................................................9 2.3.2 Continuous allocation mode..................................................................................................9 2.3.3 Interval allocation mode......................................................................................................10 2.3.4 The comparison of two allocation modes...........................................................................10 2.4 1*3 probability of ad-frequency collision .....................................................................................12 2.4.1 Distribution of ideal meshes...............................................................................................12 2.4.2 Irregular network.................................................................................................................13 2.5 1*3 reuse technique impact on network quality ...........................................................................15 2.5.1 Frequency hopping influence on speech quality................................................................15 2.5.2 Impact on 1*3 network performance caused by C/I...........................................................16 2.5.3 Impact on 1*3 caused by layout of sites.............................................................................16 2.5.4 Impact on 1*3 network performance caused by engineering parameters .........................17 2.5.5 Impact on 1*3 network capacity cause by handover..........................................................21 2.5.6 Impact on 1*3 network cause by load handover................................................................23 3 1*3 frequency-hopping data configuration...............................................................................................23
2005-01-18
Public Use
Abstract: This document combines radio network layout and application experience of 1*3
reuse. It is a guideline to introduce the principles and measures of 1*3 tight reuse frequency planning. Name Author Reference List Code Released date Where and how to access Publisher
1 General
2005-01-18
3
Public Use
Due to the shortage of frequency resource, in recent years equipment manufactories and operators have been focusing on improving the efficiency of frequency utilization in GSM system, and they try to improve network capability in limited frequency resource.
2005-01-18
Public Use
K = N W /N TRx
K is Frequency Reuse Factor; NW is the number of available frequencies; NTRX is the maximal amount of carriers in a cell. When 15 frequencies are used for the carriers participating in frequency hopping and the number of FH carriers in the cell is 2, Frequency Reuse Factor K = 7.5. When the number of FH carriers is 3, K = 5.
Figure 1 The overlapped depth of coverage is different in cells Figure 1 The coverage overlap Because the different overlapped depth of A and B network, the number of As adjacent cells is less than Bs, so the interference of B is bigger than As. Conclusion: The more adjacent cells are, the bigger the probability of co-frequencies collision is, and the lower the utilization efficiency is. Therefore, the amount of adjacent cell should be decreased whichever frequency reuse technology is used.
1.3.2 Disadvantages
1. Broadband combiner is needed and cavity combiner with the property of frequencies
2005-01-18
Public Use
selection cant be used on 1*3. 2. Broadband repeater is adopted, because 1*3 has much influence on frequency-selected repeater. 3. As the reuse distance decreases, interference of ad-frequency and co-frequency will increase rapidly. 4. Network needs delicate optimizing adjustment. Especially, The overlap of coverage should be restricted strictly. It is worthy to point out that most sites configuration can only be S2/2/2 and few of them can be S3/2/2 while we adopt general 4*3 tight reuse technology and 6MHz band is available. Otherwise the network performance will be out of control. When 1*3 close reuse technology is used, the maximal configuration is S4/3/3(but it is a theoretical value, the actual configuration is S3/3/3 generally). Moreover, the capacity is twice of 3*4 reuse technology, which can save the invests of operators greatly (the expense of tower, equipment room, power supply, transmission and other assistant equipment will be higher than the BTS equipment).
2005-01-18
Public Use
resource, maximal site configuration which could meet the requisition of capability nowadays and in the future. What is ideal mesh? The relative position of ideal meshes must meet some mathematics relation, the relation of the equilateral triangle.
Figure 2 The allocation of ideal stations The experience proves that the quality of network and utilization efficiency of frequency will be best when the sites are based on ideal meshes. In other words, more users will be contained.
2005-01-18
Public Use
Notes:The capacity is theoretical values, in actual planning 70~80 percents of theoretical capacity above is available.
Figure 3 The capacity increasing The capability will increase 99 percents after using 1*3 tight reuse technology under the condition the quality of network could be accepted.
Figure 4 1*3 tight reuse pattern While adopting 1*3 tight reuse technology, we must use Radio Frequency Hopping. There are two kinds of allocation mode of MA: continuous allocation mode and interval allocation mode. Principles of frequency hopping are expressed as follows:
MAI = fMAIO, FN, MA, HSN, N = fFNHSNN fMA, MAIO
2005-01-18
Public Use
where MAI is mobile allocation index; N is the amount of frequency in MA; FN is frame number. When FN, HSN, N of the three cells in the same site are all the same, MAI is only related to MA and MAIO. It shows that ad-frequencies collision intra-site can be controlled by planning MA and MAIO carefully. It also shows that ad-frequencies collision inter-site will be controlled when the amount of the frequencies in MA of the three cells is the same. NOTICE: It is different from the FH descriptions in GSM Protocol that the cells of intra-site share the same HSN. The reason is to avoid inter-cell ad-frequencies collision in the site. It is determined by BTS equipment that FN of different cells in the same site is the same. When the number of sites with frequency hopping is more than 63, those sites that are far apart between them can reuse HSN.
In continuous allocation mode, the maximal configuration is S3/3/3 (BCCH carrier without frequency hopping + 2 TCH carriers with frequency hopping). For bigger site configuration, the ad-frequencies collision in a cell is unavoidable.
2005-01-18
Public Use
In interval allocation, the maximal site configuration is S4/3/3, .For bigger site configuration, the ad-frequencies collision in a cell is unavoidable.
2005-01-18
10
Public Use
online. 5. When bandwidth is 6MHz, the maximal BTS configuration that continuous allocation mode supports is S3/3/3(theoretical value) and that supported by interval allocation mode is S4/3/3.
96,97,98,99,100
Continuous al location
96,99 ,102,105,108
Interval allocation
96,97,98,99,100 106,107,108,109,110
101,102,103 ,104,105
97, 100,103,106,109
96,99,102,105,108 97,100,103,106,109
106,107,108,109,110
101,102,103 ,104,105
98,101,104 ,107,110
97,100,103,106,109
Figure 5 1*3 Instance of two allocation mode Notice: In the above allocation, BCCH frequency 111 should be used as less as possible. Especially, it cant use in the third cell (the cell contained 110 in MA). When continuous and interval allocation mode are used in one actual network (idle BURST send testing), there is no difference in coverage. On the other hand, continuous allocation mode is better than interval allocation mode in receiving quality and their difference is listed below: 50
40 30 20 10 0
ua l_ ua l_ ua l_ ua l_ LE ua ua ua ID Q Q Q Q Q Q Q Q ua
2005-01-18
l_ 1
l_ 4
l_ 5
l_ 7
11
Public Use
Figure 6 1*3 quality difference of two allocation mode Figure 6 The receiving quality of two allocation modes
Figure 7 1*3 tight reuse technology In this figure, there are no co-frequency collisions in cell A-3, but there are ad-frequency collision in A-3 with B-1, D-1, D-2 and C-2. In the figure, the number of ad-frequency collision cells is listed below: Table 5 The number of cells which there are ad-frequency collisions with A-2 Interference area Interference cells number X 2 X1 2 X2 1 X3 2 X4 1
Because it have been assumed that engineering parameters and propagation environment are all the same, the receiving levels of A-3 and adjacent cells are same at the receiving points. Because the interference at X, X1 and X3 is maximal, only calculating the probability at X is enough (The frequencies number in MA is 5):
2005-01-18
12
Public Use
1. When there is one carrier participating in FH in A-3, D-1, and D-2, the maximal probability of ad-frequency collision is:
P=
1 5
1 5
1 5
1 5
2 25
= 8%
2. When there are 2 carriers participating in FH in A-3, D-1, and D-2, the maximal probability of ad-frequency collision is :
P=
2 5
2 5
2 5
2 5
8 25
= 32%
3. When there are 2 carriers participating in FH in A-3, D-1, and D-2, the maximal probability of ad-frequency collision is :
P=
3 5
3 5
3 5
3 5
18 25
= 72%
Notes: when three carriers participate in FH and the frequency number in MA is 5, continuous allocation mode can't avoid ad-frequency collision intra-cell. But interval allocation mode can avoid ad-frequency collision intra-cell. When the site configuration is lower than S4/3/3, interval allocation mode can avoid ad-frequency collision between adjacent cells of the site. When the configuration is higher than S4/3/3, interval allocation cant avoid ad-frequency collision between adjacent cells of a site. The calculation proves that when MA is fixed, probability of ad-frequency collision has direct ratio with square of carriers participating in FH. In other words, ad-frequency interference will increase rapidly with increasing of network capability. It needs to be pointed out that the calculation above is done when network runs under full load. Actually the network load is lower than the full. One connection in A-3 cell will cause ad-frequency interference to connections, which locate on the same timeslots in D-1 and D2, and there is no ad-frequency interference to the other timeslots. Therefore, ad-frequency collision of actual network is related to the number of connections.
2005-01-18
13
Public Use
2) The cell has j-2 adjacent-cells of other BTS, including k adjacent cells have the same number with the current cell. 3) The number of frequencies participating in FH is N. 1. When there is one carrier participating in FH, The probability of co-frequency collision:
P = (1 n
1 n)
k = k/n 2
P = (1 n
1 n)
(j 2 k) = (j 2 k)/n 2
2. When there are two carriers participating in FH, The probability of co-frequency collision:
P = 4k/n 2
The probability of ad-frequency collision:
P = 4(j 2 k)/n 2
3, when there are m carriers taking part in frequency hopping The probability of co-frequency collision:
P = m 2 k/n 2
The probability of ad-frequency collision:
P = m 2 (j 2 k)/n 2
The probability of collision is related to connections in network. The key points to garanntee network performance: 1. The layout of the sites locations should be distributed along the regular meshe, and antennas height should be almost same during the planning and design period. 2. In optimizing period the coverage should be controlled right to aviod the co-frequnecy adjacent cells with the same number. According to calculation listed above, C/Ia should be 0 in X area, but the actual C/I is much lower than this value because of the coverage overlap, the fast fading and handover threshold. Therefore, the key works during optimization are to reduce the depth of overlap coverage and improve the handover sensitivity.
2005-01-18
14
Public Use
2.5 1*3 reuse technique impact on network quality 2.5.1 Frequency hopping influence on speech quality
According to subject evaluation of speech quality, FH has much impact on Rx Qual of MS. Testing result and subject evaluation is listed below: Table 6 Rx Qual difference between FH and without FH 0 A A 1 A A 2 B A Rx_Qual 3 4 B C B B 5 D C 6 D D 7 E E
Subject evaluation
Without FH FH
Table 7 Subject evaluation grade Subject evaluation grade A B C D E Evaluation criterion Very clear, no noise clear, a little noise Understood, noise Understood after repeating Cant be understood
Testing result proves: Receiving quality and subject speech quality in FH is different from that in without FH. When FH is not used, Rx_Qual is 0 or 1, subject speech quality is A; Rx_Qual is below 3, subject speech quality is clear. When FH is used, Rx_Qual is 0,1 and 2, subject speech quality is A; Rx_Qual is below 4, subject speech quality is clear; Rx_Qual is above 6, there are no difference between FH and without FH. Subject evaluation and quality grade between FH and without FH is showed below:
8 7 6 Rx_Qual 5 4 3 2 1 0 3 4 5 6 subject evaluation Figure 8 The difference of Subject evaluation and quality grade 0 1 2 A A B B C C Rx_Q ual no FH Rx_Q ual FH D E
2005-01-18
15
Public Use
Figure 8 Subject evaluation and Rx_Qulity 2.5.2 Impact on 1*3 network performance caused by C/I
The drive test data of idle BURST sending (at this time interference cell send out data continuously without power control) is analyzed in detail.
The relation of C/I and receiving quality grade from the driver testing data is shown as follows:
When C/I of 1*3 network is greater than 10dB, the subject speech quality can reach B (clear, a little noise). When C/ I is between 3~10dB, subject speech quality is C (understood, noise). When C/I is less than 3dB, network performance will deteriorate rapidly. Testing data proves that interference source in which quality grade is lower than 3 is caused by ad-frequency collision between different-numbered cells of adjacent sites.
100. 00 90. 00 80. 00 70. 00 60. 00 50. 00 40. 00 30. 00 20. 00 10. 00 0. 00
d G e0 ra d G e1 ra d G e2 ra d e G ra 3 d G e4 ra d G e5 ra d G e6 ra d e 7 G ra
16
R ecei vi ng Q ual i t y
2005-01-18
Public Use
Figure 9 The comparison of receiving quality ratio between regular cells and irregular cells The data above is originated from an actual running network, and the conclusion can be made that the distribution of the sites is very important for 1*3 network.
According to data above, after 4*3 reuse network is changed to 1*3 and before optimization, SDCCH congestion rate deteriorate greatly, handover success rate drop, SDCCH call drop rate and TCH congestion rate change little and traffic increase a little. After optimization 1*3 network, each index is improved. Compared with index before optimization, five key indexes (SDCCH congestion rate, SCCH call drop rate, TCH call drop rate, TCH congestion rate and handover success rate) have exceeded or reached indexes before optimization.
2005-01-18
17
Public Use
4*3 1. 40 1. 20 1. 00 % 0. 80 0. 60 0. 40 0. 20 0. 00
1*3 w t h O i mzat i on i pt i
SD C cal l drop CH
SD C congesti on CH
TC cal l drop H
TC congest i on H
Figure 10 The traffic statistics indexes contrast 200. 00 180. 00 160. 00 140. 00 120. 00 % 100. 00 80. 00 60. 00 40. 00 20. 00 0. 00 Traf f i c H andover success 4*3 1*3 Wt hout O i mzat i on i pt i 1*3 Wt h O mzati on i pti i
2005-01-18
18
Public Use
Table 10 Other traffic statistics indexes contrast HO Traffic 4*3 1*3 before optimization 1*3 after optimization 166.79 172.93 178.87 Interf.B and 3 2.41 2.40 4.00 Interf. Band 4 0.15 0.28 0.40 Interf. Band 5 0.16 1.56 0.11 Requests of BQ 859 2433 1832 HO Requests 9985 12716 11895 Call set up 10362 11190 11935 Average HO times per connection 0.96 1.14 0.99
i nt erf . band 4
i nt erf . band 5
Figure 12 Other traffic statistics indexes contrast The number of idle channels falling into interference band 5 is much lower than that of before 1*3 optimization.
BQ handover s 14000 12000 10000 8000 6000 4000 2000 Handover r equest s Cal l set up
2005-01-18
0 4*3
19
Public Use
2005-01-18
20
Public Use
Figure 14 Other traffic statistics indexes contrast Comparing 1*3 FH network before optimization with 4*3 no FH network, handover times increased by about 2700 times and total Bad Quality handovers increased by about 1600 times. Comparing 1*3 FH network after optimization with 4*3 no FH network, handover times increased by about 1900 times and Bad Quality handovers increased by about 1000 times. Compare increase of Call set up times with increase of traffic, the conclusion can be made that traffic increasing is natural. The increase of handover times makes little contribution on traffic increasing, and it will be explained in next section. The increasing proportion of total bad quality handovers is much higher than that of traffic. On one hand, it is due to the closer reuse frequency and the irregularity of real experimental network causes interference in some area; on the other hand, interference handover threshold (50) in 4*3 reuse is equal to 1*3 tight reuse threshold. The subject speech quality of frequency hopping whose Rx_Qual is equal to 5 is as good as that of no frequency hopping network whose Rx_Qual is equal to 4(explained in 2.5.1 impact on network quality caused by frequency hopping). The difference of subject voice-quality standard is the main cause of the increase of bad quality handovers.
2005-01-18
21
Public Use
asynchronous handover have been made on GSM1800 network that havent been in use commercially. Signaling time from the new cell's CHANNEL ACTIVATION (caused by handover) to the old cell's RF CHAN RELEASE ACK is gotten by testing. Synchronous handover counts 20 times and asynchronous handover counts 11 times. The result is listed below: Table 11 The time of two channels are occupied by one connection during HO (1800M) Average time(ms) Synchronous handover Asynchronous handover times are 18. Table 12 The time of two channels are occupied by one connection during HO (900M) Handover time Testing of traffic statistics is done in laboratory. Statistic period is 15 minutes and MS handover between cell A to cell B. During testing period, there are no other subscribers using the two cells and another MS seizures cell C all the time as the terminated. Testing results are listed below: Table 13 Lab tests Cell A Cell B Add up Cell C Statistic time(min) 15 15 15 15 Handover times 9 9 18 0 Statistic traffic(Erl) 0.1400 0.1125 0.2525 0.2500 Average time(ms) 745 Shortest time (ms) 584 Longest time(ms) 1675 348 408 Shortest time(ms) 339 398 Longest time(ms) 420 450
The test of handover period is done in GSM900 network with large traffic, and handover
Only one subscriber occupied the channel in cell A or B during 15 minutes testing period, If no handover occurs, the traffic should be same between them. But the total traffic of Cell A and B is 0.2525Erl, and the traffic of cell C is 0.25Erl. The excessive 0.0025Erl traffic is due to handover, and the average crossed-time every handover is 0.5 seconds. Notice: because the shortest interval of traffic statistics is 480 ms, handover crossed-time calculated by traffic statistics has a certain error. Comparing network with large traffic and low traffic, handover crossed time increases greatly while traffic become high. It is assumed average handover crossed-time is 0.78s, and the increasing traffic is 2Erl every 10000 handover (0.75*10000/3600).
2005-01-18
22
Public Use
Comparing the handover increase of 1*3 network before optimization with that after optimization, suppositional added traffic is about 0.4Erl caused by 2000 handovers. So 12Erl traffic increasing is natural. Conclusion: Suppositional added traffic is about 2Erl when 10000 handovers are made. Impact on traffic caused by suppositional added traffic should be considered when handover times are extremely high. On the other hand, data above proves that handover speed is related to traffic (the more traffic, the longer handover time).
Comparing traffic statistics indexes between enabling and disabling load handover, TCH congestion rate is not high, but it hasnt been lowered greatly after enabling load handover and call drop rate increases distinctly. Conclusion: Load handover cant be employed in 1*3 network. For example the bandwidth of load handover is 25dB and the connections meet the load HO conditions, they will handover to second best cell and seize FH channels with serious interference, and call drop rate will increase distinctly. In order to ensure that MS camp in cells with strongest signal, cell selection and reselection parameters should be consistent with each other in 1*3 network.
2005-01-18
23
Public Use
Table 15 1*3 data configuration related to frequency hopping Menu name Table name Parameter FH index No. Radio Channel Configuration Table MAIO Local office FH index No. Frequency Hopping Data Table TSC ARFCN 1~ ARFCN N ARFCN 1~ Carrier Site Configuration Table Cell Configuration Cell Data Table Cell Allocation Table FH mode ARFCN 1~ ARFCN N ARFCN N Static TRX Power class HSN 0~1023 0~N-1 Value 0~1023 Annotation Index to frequency- hopping data table. And the value of the FH TRX carriers in a cell should be same. Mobile Allocation Index Offset. In this case, the same MAIO is recommended for all channels of a TRX and different MAIO for different TRX in the same cell. Correspond to item in Radio channel configuration table Hopping Sequence Number. HSN in different cells of the same site is the same Training Serial Code, Same with BCC. Frequencies in MA participating in FH BCCH frequency and frequencies in MA participating in FH Power class "0" shows that power is in its maximum. Each class is 2dB less than its former class. FH mode should be RF FH for 1*3 frequency reuse pattern BCCH frequency and frequencies in MA participating in FH
1~63
0~7 Available frequency Available frequency 0~10, unit:2dB Radio frequency hopping Available frequency
2005-01-18
24