The impact of CPICH power variations and downtilt angles on UMTS based systems has been studied in this chapter. The study goes beyond well known facts and interactions with other parameters are also investigated.
The study starts by analyzing the CPICH. Its participation in the cell selection and SHO procedures determine its preliminary adjustment by means of standard link budgets. In this sense, it has been shown an analytical example providing a closed expression to estimate the required value.
By means of simulations (in a real and in a synthetic scenario), it has been quantified the transfer of UEs by using CPICH power variations, the technique is effective to reshape cell areas and even small changes have a noticeable impact.
Regarding interactions with SHO, although the number of DL connections are directly proportional to the value of AS parameters N and AddWin, paradoxically the number of UL connections is smaller because UEs have more possibilities to transmit to better option cells. From this, it has been observed that higher SHO parameters help to reduce the impact of wrongly adjusted, or just non-optimized, pilot powers.
Once SHO parameters are fixed, it has been quantified how AS sizes are changed by CPICH power variations. Although the scenario has a quite heterogeneous distri- bution of UEs, variations are present but are not very significant, and SHO overhead is maintained. Hence, only uncoordinated changes of pilot powers could jeopardize this aspect.
UL, the maxim that dictates that transferring UEs to other cells contributes to reduce the own load must be taken carefully. There is an optimum pilot adjustment that implies a proper assignment of UEs into cells and once these values are left, interference increases, regardless the pilot is increased or decreased. Even faraway cells are impacted by a wrongly adjusted pilot and so the whole area to optimize must be studied as a whole.
Regarding DL load variations, after comparing two definitions on the DL load factor, the analysis shows that there exists a monotonous increase with the pilot level, reinforced by the extra power consumption of control channels. On the other hand, reducing the CPICH power to unload the DL of a particular cell is effective but at the cost of degrading nearby ones. As seen in the previous chapter, many published works perform DL based balancing in scenarios with one hotspot. If the scenario has a more heterogeneous distribution this rule of thumb must be taken carefully and the whole area to optimize should be studied simultaneously. This is one of the requisites in the designed proposal in next chapter.
The impact of pilot power variations on system capacity represents the second part of this study and its main conclusions are summarized next:
• Optimum CPICH values are not very dependant on AS modifications. • Higher SHO parameters cause a less sensitive network to wrong adjustments in
CPICH powers and capacity degradation is less relevant. HHO requires a finer adjust of power and small variations cause important capacity reductions. • Conversely, a correct setting of pilot powers promotes reducing the differences
in capacity among different SHO parameters. So it contributes to minimize the capacity reduction when these are not optimally adjusted.
• DL capacity is more sensitive than the UL to increases in pilot powers be- yond the optimum value. Reductions imply far less capacity degradation but coverage issues must be considered.
• By properly adjusting CPICH powers and AS parameters, capacity balancing between UL and DL can be achieved.
Downtilt of antennas is the second planning parameter that has been investi- gated. This technique appears to be particularly appropriate for 3G networks using a unitary frequency reuse, the reason is two-fold: first to increase isolation among cells and second because of the recent technological evolution that permits remotely controlled electrical tilt through new interfaces such as Iuant from 3GPP.
Coverage and interference variations caused by downtilt variations have been previously addressed in the literature. Along this chapter, new detailed results are given in the framework of UMTS systems and its performance optimization. Dif- ferent performance indicators have been evaluated for different electrical downtilt variations in a scenario of variable load and the main conclusions are listed next:
• Not only the main lobe slope down is important but also the relative position of the first radiation null and secondary lobes with respect to UEs from limiting cells.
• From the previous point, a correct minimization of intercell interference leads to the optimum angle in terms of UL load factor, considered as the optimum from the whole cell viewpoint.
• However individual UL transmission powers depend on both the noise rise and the link loss. It is the global effect what determines the final required power. For this reason, in general, the optimum angle from UEs viewpoint does not match with the optimum from a cell viewpoint. Actually, both configuration are only the same for high loaded scenarios.
• Adjusting angles to reduce required UL transmission powers permits increas- ing the UE perceived QoS but at the cost of being in a more instable situation. Slight changes in the system load imply important power increases and the probability of having UEs in degraded mode grows rapidly. On the other hand, minimizing the load factor yields a more steady system but at the cost of requiring higher levels of power when loads are not particularly high. • DL effects are simpler to analyze and conclusions on the optimum angles
match with the UL load study.
Given the previous points, in order to design a static Automatic Planning algo- rithm, angles should be optimized from a cell viewpoint, so that both UL and DL performs correctly under variable load conditions. However, by introducing remote and electrically controlled antennas, angles could be dynamically modified accord- ing to traffic patterns mapped to a timetable. In that case, whenever the UL is the limiting link, optimal angles from UEs viewpoint could be tracked and achieve the best possible network performance.
The worst possible downtilt adjustment implies having the first radiation null pointing towards the cell and the second lobe pointing towards the adjacent ones. This is an important issue in Dynamic Planning strategies that change cell sizes by only adjusting CPICH powers and which do not consider the spatial distribution of UEs after having reshaped the cell. In fact final figures show that optimum angles are very dependant on cell sizes and important capacity degradations are obtained if downtilts are not accordingly adjusted.
As a final comment, it is worth mentioning that modifications in both pilot and downtilt permit a controlled way of changing the cell shape. Nevertheless their effect is noticeable different. Pilot variations just imply a change in the power assigned to control signals and also have an influence on the reliability of UEs channel estima- tions. On the other hand, downtilt variations have a direct impact upon the power UEs and BSs have to transmit, the radiated interference and also interference levels listened from other cells. In this sense, using CPICH power variations to achieve UEs transfer among cells looks more appealing than downtilt variations, that appears to be much more indicate to control transmission powers and interference.
Automatic Planning Based
on Simulated Annealing
4.1
Introduction
Along the previous chapter it was analyzed the influence of the CPICH power levels and downtilt of antennas on the capacity of the network. It was shown that variations on these parameters play a major role on the optimization of the radio access network and several guidelines for their adjustment and the reduction of load levels were devised.
The use of equal pilot powers and downtilt angles would be the best solution in terms of interference in an ideal scenario with perfectly uniform user distribution, equal path loss conditions and only one service, which imply homogeneous traffic among all cells. Although this design can be intuitively predicted, the idea is ana- lytically addressed in [AHNPM01], which demonstrates that this uniform network layout with equal-sized cells is optimal for that theoretical case, and that capacity degrades significantly if loads are not balanced.
Unbalanced traffic distributions in realistic environments might lead some groups of cells to congestion whereas other ones could be remaining with a much lower load. This unbalance can be produced by non-uniform users distribution and by non-uniform services distribution. In this sense, from the previous chapter it is clear that variations in CPICH powers and downtilt angles are potentially good strategies to improve capacity by reshaping effective cell areas. In particular, it can be predicted the existence of a group of CPICH powers and downtilt angles such as traffic is effectively equalized among cells and a higher capacity is achieved.
Along this chapter an Automatic Planning strategy is designed and assessed to adjust both parameters. From the study in the previous chapter, it is necessary to optimize jointly all the target cells and to consider realistically UL requirements.
Indeed this is one of the novelties of the proposal since this link is usually missed in the literature, as was pointed in Chapter2.
After this introduction, the chapter states the problem analytically and stresses the importance of correctly assigning cells to UEs in the UL. Next section gathers the principles of the proposed solution and fully describes the Automatic Planning strategy. The fourth section presents results and the discussion. Finally the chapter is closed by conclusions.