Published on

Cell Breating

Published in: Technology
  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

  1. 1. Cell Breathing in CDMA Networks Angela Amphawan Abstract EbA EbBThe widespread deployment of wireless cellular = (1)technology affects the traditional concept of channel N tA N tBallocation and drives the necessity for novel load where Ebi/Nti is the signal to noise ratio received at basebalancing techniques. Various load-balancing schemes station i for the mobile under consideration, Ebi is thehave been proposed to assign channels to the cells such received bit energy and Nti is the spectral density of totalthat the available channels are efficiently used and thus interference at base station i.channel reuse, maximized. This is particularly importantin networks where traffic distribution is non-uniform. RSome of the well-known load-balancing schemes includethe Directed Retry (DR), Channel Borrowing (CB), Cell CellChannel Borrowing without Locking (CBWL), Load A B XBalancing without selective borrowing (LBSB), Cell Forward andBreathing and a number of others. Each possesses both reverse linkvirtues and drawbacks. The objective of the paper is to handoff boundaryprovide an assessment of the capabilities of Cell Soft handoff regionBreathing in optimal CDMA network management. Thepaper studies the capabilities of cell breathing, its flaws Figure 1. A balanced forward and reverse linkand suggests possible solutions to the predicament. handoff boundary case for two cells A and B Forward link handoff boundary is defined as the1. Introduction contour of the mobile locations where E cA EcB Cellular communication networks divide the = (2)geographical area into smaller hexagonal regions, called Io Iocells. Over the last few years, a great deal of work has where Eci is the received pilot chip energy of i-th pilot andbeen done to study optimal CDMA cell design whereby Io is the spectral density of the total power seen by thethe number of cell sites required is minimized while mobile.maintaining the quality of service provided. The focus of It can be seen from (1) and (2) that, if the interferencethese studies was on the maximization of cell radius for a levels are the same at both base stations and the samegiven maximum power transmitted while ensuring that amount of power is transmitted on the pilot channel fromthe quality of service in a specified percentage of the cell each base station, then the forward and reverse handoffarea was met. Most studies were under the presumption of boundaries will coincide; the boundary will be half wayuniform traffic distribution in each cell. between the two cell sites for a uniform propagation In this paper, we analyze the impact of non-uniform model.traffic distribution in different cells on cellular network As the reverse link traffic load is increased, the thermalperformance. Specifically, we are interested in noise at the base station increases. It is clear from (1) thatunderstanding how cell breathing performs load balancing the reverse link handoff boundary will move closer to theunder non-uniform traffic distribution. base station whose rise over thermal noise is greater. Cell breathing is a mechanism that attempts to keep the Then, to balance the reverse and forward link handoffforward and reverse link handoff boundaries balanced by boundaries, the pilot signal of the cell with greater basechanging the forward link coverage according to the station interference must be reduced. The mechanismchanges in the reverse link interference level. used to reduce pilot power on a cell based on the increase Reverse link handoff boundary is defined as the in the reverse link interference level is referred to as cellcontour of mobile locations between neighboring cells breathing.where the received signal to noise ratio at the two base A brief overview of the soft handoff process is given instations is the same. Referring to Figure 1, the reverse link Section 2. Positive impact of breathing on the forwardhandoff boundary between cell sites A and B is the link is discussed in Section 3 while Section 4 discusses itslocations such that
  2. 2. negative impact. In Section 5, possible solutions are On the forward link, it is generally believed that once asuggested. The paper is concluded in Section 6. given cell becomes heavily loaded, then the breathing2. Overview of the Soft Hand over Process algorithm will reduce the cell’s coverage thereby shedding some of the traffic to the surrounding cells As discussed in Section 1, in IS95 CDMA systems the relieving the overloaded cell. Shedding of traffic from themobile measures the pilot Ec/Io from neighboring cell heavily loaded cell occurs because, as discussed insites. If a pilot is found whose Ec/Io is above a threshold Section 2, the soft handoff region on the side of the cellcalled T_ADD, the mobile reports that pilot to the base with less loading shrinks. As a result, some of thestation. The pilot is to be included in the set of pilots in mobiles in the lightly loaded cell which were in softsoft handoff referred to as the active set. On the other handoff with the heavily loaded cell will now go out ofhand, if the Eci/Io of a pilot in the active set is below a soft handoff with the heavily loaded cell. This releasesthreshold called T_DROP for more than a certain time, some of the radio links from the heavily loaded cell thatthe mobile will report that pilot to the base station. The serviced the users in soft handoff from the neighboringpilot may then be removed from the active set. Therefore, cells. The released radio links can be used to support newthe pilot Ec/Io values as measured by the mobile users in the heavily loaded cell. The question is howprimarily determines the handoff region. Figure 1 shows much capacity increase may be expected in the heavilythe handoff boundary of two adjacent cell sites marked as loaded cell and what the impact will be to the overall cellA and B. Since Io = IoA + IoB, where Ioi is the power quality.spectral density of the total signal received from cell site i In a situation where one specific cell is heavily loadedat the mobile and No is the thermal noise power spectral and all its surrounding cells are lightly loaded, ifdensity, then we get Io ≈ IoA near the edge of the soft breathing algorithm is activated then the handoffhandoff region closer to cell site A. In other words, the boundary of the heavily loaded cell will move furtheredge of the soft handoff region near one cell site is inside that cell. In other words, the soft handoff boundaryprimarily determined by the total signal power from that will move closer to the heavily loaded cell, Figure 2.cell site. Suppose that cell A in figure 2 is heavily loaded and cell Since the left side of the soft handoff region in Figure B lightly loaded. Then, in order to move the handoff1 is determined by EcB/Io, i.e. the signal to interference boundary by an equal amount that the interference hasratio seen on pilot B, then the left side of the handoff risen in a cell, the breathing algorithm will introduce anregion does not move when the pilot power of cell A is attenuation equal to α on the forward link of cell A inreduced. The right edge of the soft handoff region response to a rise over thermal noise in cell A’s reverse(Figure 1) however is determined by EcA/Io; therefore, as link. The pilot by EcA/Io seen from pilot of cell A will bethe pilot power of cell A is reduced, the right edge of the changed tosoft handoff region moves closer to cell site A as shown αEcAin Figure 2. Therefore, if cell breathing is active, then as (3)cell loading is increased in cell A and surrounding cell I oA + I oB + I oCsites remain lightly loaded, the soft handoff region inside which is smaller than it used to be prior to breathing. Thethe neighboring lightly loaded cell sites will reduce. Of α factor is due to breathing on cell A. Now the edge ofcourse, if the loading in all cell sites increase uniformly the soft handoff region closer to cell B will move awaythen the breathing algorithm will reduce the soft handoff from cell B. Assuming that 35% of the cell area is in softregion by roughly equal amount in all cells. In the next handoff and approximating the cell area by a circle, wetwo sections, we investigate the impact of cell breathing have X=0.8R in Figure 1. We note that for large path losson network performance. exponents of 4, inside cell B the denominator of (3) will Cell R Cell be dominated by IoB. Therefore, the handoff boundary A B moves to the left in Figure 2 approximately 1 dB for each dB of attenuation introduced by breathing on cell A. As the handoff boundary is moved closer to cell A, the X Forward and reverse handoff one benefit that may be obtained is that users that are boundaries inside cell B and are in soft handoff with cell A may fall after breathing out of soft handoff with cell A, relieving some capacity from cell A to be used for users that are inside cell A.Figure 2. Handoff boundary has moved closer to However, the attenuation α is applied to the total power cell A due to loading on cell A going out of cell A, the overhead as well as the traffic channels; therefore, we need to assess the impact of this3. Positive Impact of Breathing on Forward attenuation on the traffic channels in cell A. The SNR Link Performance seen by a mobile prior to attenuation introduced by breathing is given by:
  3. 3. L gI or β j setting them equal we find that g’=1.1g. In other words, SNR = ∑ (4) there is a 10% increase in forward gain in about 54% of j =1 (1 − β ) I or + I oC + N o the cell area. Note that this 54% of the mobileswhere g is the fraction of total forward link power given correspond to the outer area of the cell. If we let Ioc/Iorto a mobile referred to as the forward gain, L is the total =0.5, which corresponds to about 40% of the cell area,number of multipath components captured by the mobile then substitution into (4) and (5) gives g’=1.14g.from cell A, Ior is the total power received by the mobile Therefore, as mentioned above, the forward gains willfrom cell A, βj is the fraction of the total power received increase for all mobiles as breathing introducesby the mobile from cell A that is received on the j-th attenuation.multipath component (which sum to 1 over all multipath In the two-path model described above, the increase inj), IoC is the out of cell interference. Note that the first the forward gain for the traffic channels is initially lesscomponent in the denominator of (4) is due to in-cell than the breathing attenuation, but as the forward gainsinterference, from the multipath. Once the breathing are increased initially, the total power going out of theattenuation is applied, the SNR seen by the mobile at the cell increases. Then, we can apply equation (5) iterativelysame location is given by to find out how much the forward gains will be increased L λg ′I or β j in response to the increase of in cell interference. SNR = ∑ (5) Therefore, eventually the traffic channel forward gains j =1 (1 − β ) I or + I oC + N o will increase by the same amount as the breathing The difference between (4) and (5) is in the attenuation attenuation, until the total traffic channel power going outα and g’. In order to achieve the same SNR for the mobile of the cell becomes close to what it was prior to breathingafter breathing as before breathing, we need to increase attenuation. The output power after breathing will be lessg’, the fraction of power allocated to the mobile. In other due to lower power on overhead channels. Note that wewords, we need to determine g’ such that (4) and (5) are are assuming that the overhead channels (pilot, pagingequal in order to maintain the same FER. Therefore, the and synchronization) are not power controlled and theirforward gain of the traffic channel will increase to power is therefore reduced due to the breathingachieve the same SNR as before. Note that in the actual attenuation. The reduction of total transmit power due tosystem the reduction in the SNR due to breathing reduced power on the overhead channels will eventuallyattenuation will cause the FER o f the users to increase. be used by new users and therefore the total transmitThe power control will then automatically increase the power will remain unchanged before and after breathingforward gain of all users whose FER has increased in under heavy loading conditions. Of course, the mobileorder to lower their FER values to the desired values. whose forward gains were near their upper limit, their If there is a single path seen by the mobile, it is then allocated power will decrease due to hitting the upperclear from equations (4) and (5) that g’ must increase by limit of the forward gain. Recall that based on the discussion in Section (2), thean amount equal to α in order to maintain the SNR. edge of the handoff region on the cell site i is determinedSuppose α=1dB. Then the soft handoff region will move by the ratio Eci/Io. Based on the above discussion, theby about 5% for the path loss exponent of 4 which means power being transmitted from the heavily loaded cell sitenow about 25% of the area of cells surrounding cell A will remain almost constant after breathing. Then, thewill be in soft handoff with cell A instead of 35% of the edge of the soft handoff region inside cell site A will notarea prior to breathing. In other words, 10% of mobiles in move much because it is determined by Eci /Io which hascommunication with cell A may be shed to other cells by not changed. The edge of the soft handoff region insideputting them out of soft handoff with cell A. Note that cell B which is determined by EcA/Io, however is reducedhere we are assuming that all surrounding cells are lightly because the pilot power transmitted from cell A isloaded resulting in a net 10% shedding of traffic to the reduced due to breathing. The reduction of soft handoffsurrounding cells. Throughout, we have assumed circular area in cell B results in the reduction in the number of softcells and that cell A is surrounded by 6 other cells. Figure handoff links that cell A must support for users inside cell1 and 2 only show 2 cells for simplicity. B. Therefore, the main sources of capacity increase in the In a single path case, it is clear from equations (4) and scenario where a heavily loaded cell is surrounded by a(5) that there needs to be a dB for dB increase of forward tier of lightly loaded cells is the shedding of traffic togain for each dB of breathing attenuation in order to other cells and lower power on the overhead channels onmaintain the traffic channel SNR. the heavily loaded cell. There will also be some Next, consider the case where there are two multipath additional capacity due to the mobiles, which were at theircomponents with the weaker paths’ power equal to half upper limit of forward traffic gain; these mobiles’ powerthe power of the stronger path, i.e. β1=0.67 and β2=0.33. cannot be increased which leave some room for newAlso, assume Ioc/Ior=0.25; separate simulations have traffic. But this additional capacity is at the cost of highershown that in about 54% of the cell area Ioc/Ior>0.25. In FER for some mobiles.this case, by substituting into equations (4) and (5) and
  4. 4. Based on the above discussion, the forward link increases, thus deteriorating the cellular networkcapacity of a heavily loaded cell, which is surrounded, by performance.lightly loaded cells may be increased through cellbreathing. Cell breathing increases the capacity of the 5. Solutionsheavily loaded cell using two mechanisms. First, cellbreathing reduces power on the overhead channels equal Based on the above discussion breathing may provideto the amount of breathing attenuation. For instance, if a small amount of capacity gain in very limited traffic25% of the power had been allocated to the overhead scenarios. However, there may be an impact to callchannels and 1dB attenuation was applied to cell, then the quality if excessive breathing is allowed. The amount ofamount of power on overhead channels would reduce to breathing attenuation must be limited to a small amount20% of total available power before breathing. This so as not to adversely impact the soft handoff region andresults in approximately 6% increase in forward link overall call quality.capacity due to reduced overhead. The second The extent of cell breathing can be limited through themechanism that increases forward link capacity is by use of call admission control (CAC). The CACshedding mobiles to other lightly loaded cells. As mechanism is used to decide when a new call can bediscussed above, one dB attenuation will result in accepted. Schemes that are based on the measured noisereduction of soft handoff region of the surrounding cells rise can be used to set the minimum cell size. Any newfrom 35% to 25%. Therefore, under uniformly distributed calls will be blocked once the interference reaches atraffic conditions there may be up to 10% increase in certain level.capacity. Therefore, one dB of breathing may provide at From this, we can see that it is important to considermost 15% capacity increase in a heavily loaded cell that is both network coverage and call blocking when planning asurrounded by lightly loaded cell. CDMA network. Detailed network planning requires a specialized CDMA planning tool. These tools use traffic4. Negative Impact of Breathing on Forward information and propagation predictions to determine the Link Performance coverage of the CDMA network. Most 3G network planning tools utilize the ‘power This capacity increase may, however, be at a cost of control loop’ method. Analytical techniques can also bereduction in call reliability of calls because the soft used, but these are slow when the traffic load is high. It ishandoff region has been reduced. The reduced soft important for the tool to consider call blocking as well ashandoff may result in an increase in drop call rate. The network coverage. The tools can also be used to predictreduced soft handoff region by breathing is particularly the soft-hand over regions.problematic in networks whose soft handoff region hasalready been reduced and optimized. 6. ConclusionLightly loaded cells The impact of cell breathing on the coverage of CDMA cells was studied. It was shown that the main Heavily loaded cell impact of cell breathing is reduced soft handoff region. There is almost one dB reduction in soft handoff region for each dB of breathing attenuation. Therefore, if cell breathing is used to distribute traffic in certain deployment/traffic scenarios, the impact of reduced soft Lightly loaded cells handoff region on call quality needs to be assessed. It Figure 3a. No coverage holes develop when was shown that under very specific deployment lightly loaded cells surround a heavily loaded conditions where one cell is heavily loaded and is cell surrounded by lightly loaded cells, cell breathing might provide a small increase in capacity. This capacity gain may, however, be at some call quality cost. In order to minimize impact to call quality, the amount of breathing attenuation must be limited and controlled carefully. Coverage holesFigure 3b. Coverage holes develop when all cells References are heavily loaded 1. C. Wheatly, “Trading coverage for capacity inAlso, if situation arises where all cells are heavily loaded, cellular systems: A systems perspective,all cells will be simultaneously attenuated by the “Microwave Jl, 38 (7): 62-76, July 1995breathing algorithm, causing coverage holes to develop 2. Sajal K. Das, Prathima Agrawal, “A Distributed(Figures 3a and 3b). The probability of call dropping Load Balancing Algorithm for the Hot Cell
  5. 5. Problem in Cellular Mobile Networks”, 1997 IEEE3. A. Jalali, “On Cell Breathing in CDMA Networks”, IEEE 19984. Zhang Youngbing, “A New Adaptive Channel Assignment Algorithm in Cellular Mobile Systems”, 1999 IEEE