Looking for peak network quality

[Abstract] the quality of the wireless network and the many factors related to the complexity of the resulting network optimization work. Proposed key focus is to enhance the network quality assessment balance and focus on the discussion of the using chameleon algorithm reasonably precise configuration of power, control interference distribution network to reach peak network quality.

[Keywords:] C / I Chameleon algorithm adaptive convergence feedback 1 C / I outlined the basic criteria of 1.1 the most fundamental network quality to evaluate quality of wireless networks is the C / I (carrier to interference ratio C / improvement of I can reduce the misuse bit rate, frame loss rate, improve voice quality, MOS-aware wireless network optimization underlying indicators, while others, such as dropped call rate, handover success rate, wireless access KPI targets are in the C / I performance high-level reaction. shows, the focus of the network optimization is to optimize the C / I, deviate from this goal, the blind pursuit of high-level KPIs are unrealistic.

1.2 key to improve network quality assessment With the development of increasingly complex networks, all kinds of interference more and more, the means of the vast majority of network optimization to resolve the interference in the power configuration issue is to increase the power to anti-interference, high high interference high level qualitative difference ratio increases, this will lead to high quality of the whole network, and because of the uneven distribution of the network quality, easy to make end users produce perceptual differences, triggering complaints rise and decline in satisfaction. competitors in the competition in a dominant position, and try to keep the net throughout the same sex is equally important to network optimization is an art of balance, balance is the key to improve network quality evaluation.

For example: road test to optimize road test evaluation indicators has increasingly become the focus of the operators, optimization also naturally will pour more resources on the road tilt coverage cell frequency, high-power guarantee C / I, to enhance all kinds of road test indicators. At first glance, the effect of this practice is good, can reflect on the question came, road test road accounted for the overall covering an area of ​​less than 10%, less than 20% of users, while the strengthening of these regional C is bound to result in interference rise of 90% area and 80% of the users, will significantly reduce the C / I originally covering restricted areas affecting customer perception, can be seen from the ratio of the actual complaint, vast majority of indoor complaint ratio. therefore non-discriminatory (the overall network of roads and indoor optimization strategy can better improve the user perception, improve satisfaction, which is a balanced view.

1.3 C / I and power relations to improve the C / I is two: C or lower I. enhance user perception can also use some of the new features to improve the performance of the same C / I case of MOS, such as AMR, hopping and so on.

(1 improve the C major increase power to adjust coverage, optimizing the configuration of adjacent areas,

(2 I reduce the main method is to reduce the power to adjust coverage to optimize the frequency allocation, control of external interference.
The power adjustment most tangled, increased power, increased C and I, to increase the anti-interference ability of a single cell at the same time increase the global interference, reduce power reduce I also reduced C.

I by the white noise, external interference (of all kinds of interference and internal interference (frequency interference and intermodulation interference, visible in the adjustment of the power of the main effects of internal interference. Following simulated a C / I vs Lev curve to be a simple schematic diagram shown in Figure 1:
Wherein the horizontal axis is the average reception level of the whole network, and the ordinate for the average whole network C / I. Figure 1 is divided into three areas, a point, the specific description is as follows:
Area A: here is a low power region, the interference by the noise floor, the external disturbance intensity decision Therefore, with the enhancement of the level power, C / I rise quickly.

Area B: As the level rises, the main component of the interference into the interference within the network, due to the rise of the level of power increase of nearly C increases and I, C / I stabilized.

Area C: in depth coverage of the network, or the level of intensity of the wide coverage area has been unable to further enhance the level of the whole network upgrade is part of the district to enhance the contribution, in terms of one cell coverage area is much smaller than the master interference area, so in the case of not enhance the power of the whole network, C the rate of increase will be slower than the rate of increase of I, resulting in the C / I is decreased, while with the increase in the average level, the power-restricted area are also increasing. I increase in faster and faster, C / I accelerated decline in most of the operation of the network in Zone C (to violence with violence the result.

P point: some network state power and C / I the best balance point at the left of this point, due to lack of power, carrier to interference than not, at this point, to the right, interfere with the rate of increase rose, contained dry decline. [thesis net]
For different networks have different C / the I VS the Lev response curve, the same network in different traffic, external interference distribution case, the response curve is different, overlay adjustments, the optimal frequency is the important means of network optimization improve curve characteristics the good curve characteristics should have a higher point P, and the downward trend of Area C, more moderate response curve is determined, an important work to optimize content network characteristics (through the rational allocation of network power level, so that the network quality and run to the top position (P point.

Links to free papers Download Center http://eng.hi138.com
2 algorithm to find the best network operating point P - chameleon 2.1 graphics conversion from the above, most of the operation of the network in Zone C, not the network's best run described below are not used in the area, the network the P point run to get the best quality. then how to determine the point P? First network power level adjustment mainly depend on the maximum power level settings (affects coverage, traffic distribution parameter settings and power control (power control range, power control interval. following difficult to run the network point P:

(1 from the existing GSM network statistics, the lack of statistics on the C / I can not be a good evaluation of the effect of adjustment

(2 power level of statistical average interval value set, parameter settings and network response is difficult to achieve consistent, the abscissa position is more difficult to determine.

Need to think of an alternative method of these two indicators easy to get, easy to measure and relevance of strong indicators in the GSM network by the C / I affect the largest and most direct indicator is quality statistics of quality is also very easy to Get, this is the first step of chameleon algorithm will be above the C / I VS LEV map is converted to the actual quality of the target quality VS diagram, shown in Figure 2:
Where the horizontal axis is the target quality (from left to right, from good to bad, the vertical axis is the actual quality (bottom-up, from good to bad.

Area A: When the target lower quality requirements, the level configuration lower, C / I is relatively poor, there is still power and enhance the quality of space.

Area C: when the target requires high quality, the level configuration increases, the interference generated increases, the C / I deteriorates actual quality worse than the target value.

P line: different network characteristic optimum point P here can achieve best actual quality. Performance in Figure 2 as the target quality and achieve a consistent quality, i.e. the slope of 1, converging P line.

This graphic conversion, put the problem to simplify to the objective quality of the optimization process, the quality of statistics is very convenient, easy to adjust and evaluation work.

2.2 approximation P points to complete the graphic conversion, you still need to find the location of the point P, summed up a set of algorithms through practice and research, because of its sensitivity to capture the surrounding environment changes, adaptive adjust the wireless parameters, the network is in the best point P run, reduce interference and improve the quality of the network, hence the name "chameleon".

Chameleon algorithm The second step is to find the point P the infinite approximation method, described as follows:

Suppose the network initially target quality setting is q1, network feedback actual quality is q2, followed q2 is the target quality, the network feedback from the Q3, q3 drawn Q4 .... Since the slope of the line P is 1, so the network feedback will gradually converge, so that the network operating point on the infinite approximation P line, to achieve the best quality of the network. excluded in this process of artificial guess on the parameter setting, wholly determined by the network feedback parameter settings, the formation of the adaptive process, That network requires much energy, and will go to request access through this process, greatly improving the accuracy of the wireless parameter setting, as shown in Figure 3:
2.3 Chameleon algorithm parameter settings GSM the manufacturers power control algorithm ideal goal is to find quality conditions allowed power as low as possible, which requires a stable region of this quality and level most optimizing personnel will be given in accordance with the experience sets of interval parameters on the run now online, meticulous point may consider sub scene set the parameters of the power control what settings to suit the scene truly meet, planning, traffic, external interference, and other variables caused by changes in power demand it?
Chameleon algorithm of the third step is the bull's-eye diagram mass intervals and level interval as far as overlapping, so that power stabilized, and as much as possible to reduce the transmission power, to reduce the interference of the entire network. Doing level and quality can make reasonable conversion, because the quality parameters in the GSM network is set to 0 to 7, the level is set to-110dBm to-47dBm, apparently more sophisticated level setting mass intervals and level interval near the benefits are not two of those who deviate from the lead to the uncertainty of the power control direction.

By collecting traffic statistics data collection measurement reports, the level of quality dimensional finishing, the completion of the collection of environmental data in a cell case as shown in Table 1 below:
Among them, fill part of the figures said the proportion of sampling points below the corresponding level grade, under a level above the level of the corresponding uplink quality grade coloring plaid adding 100 through quality and comprehensive analysis of the level easily determine the stable region, to achieve the quality and level of equivalent conversion.

Can be seen from Table 1, the electric level when a lower proportion of poor quality, and the higher level weak the purpose of power control is to ensure the quality of the case as much as possible to reduce the transmit power, that is a reasonable set of power control electric interval level, should not be set too low cause poor quality, and should not be set too high interference. feedback results further correction setting accuracy.

The measurement report can clearly understand the distribution of interference condition, accurate power allocation, can be said that the number of the cell there are that many scenes, to achieve local conditions. More refined approach is the setting of the power control parameters according to different periods, in order to adapt to the different periods distribution of the traffic and interference with the times.

Is set in the parameter settings for each cell to the average value of the network level or regional level, so that can be well done the isotropic equalizer work In addition, the uplink and downlink interference distribution. not need to be calculated.

Chameleon the algorithm eventually reach the effect will be to reduce interference, very little wasted power network operation to the location of the point P overall indicator enhance.

Chameleon 2.4 algorithm in the 3G network chameleon algorithm for CDMA or 3G network implementation easier, simply BLER converges to obtain P-point run effect. Different business BLER (Block Error Ratio, block error rate of the statistical value of the existing network as the target set to run at their best, after 3 to 4 times a convergence after network adaptive feedback.

3 the chameleon algorithm effect since August 2011, Chameleon algorithm succession in various cities in the GSM network of multi-vendor equipment to use, and has achieved good results. Specific example, as shown in Table 2:
Effect from the overall point of view, the suppression of the interference within the network, TCH / SDCCH assignment success rate ,0-5 quality, call drop rate, handover success rate has improved significantly, strong adaptability to unexpected external interference and self-healing capabilities, you can quickly adjust the parameter configuration policies, stable network indicators, is a veritable chameleon.

4 Summary In summary, the Chameleon algorithm is an adaptive optimization algorithm parameter configuration based on network the interference environment analysis and feedback, currently applicable to 2G and 3G networks from the application practice, chameleon algorithm to support multi-vendor and adapt more environment ripe universal algorithm, the noise floor to reduce the network to enhance the network quality to provide a great boost, together with its operability extremely worth widely used.

Links to free papers Download Center http://eng.hi138.com

Computer Network Papers