A comparative study of deployment options, capacity and cost structure for macrocellular and femtocell networks

A comparative study of deployment options, capacity and cost structure for macrocellular and femtocell networks Jan Markendahl and Östen Mäkitalo Wireless@KTH, Royal Institute of Technology Stockholm, Sweden janmar@kth.se, ostenm@kth.se Abstract In this paper we will compare the cost & capacity performance of femtocell and macrocellular networks. The motivation is the possibility to use femtocells as complement or as replacement of wide area networks and hence to save investments in macrocell networks. In this study the femtocells thus are used as a tool for operators to reduce network costs for mobile broadband. This represents another business case than the often presented cases with focus on improved indoor coverage in homes with focus on voice services. Our techno-economic analysis is made as a comparative case study where capacity and cost is analyzed for wireless broadband deployed in a newly built office area with high user density. In addition to wall penetration losses we take into account; the level of user demand, the density of existing macro base station sites, the recent improvements in cost and spectral efficiency for radio access technologies, and the use of wider system bandwidths. The main finding of the study is that femtocell solutions, for the considered demand levels, are more cost efficient when new macro base station sites need to be deployed, otherwise macrocell solutions are more cost-efficient. Keywords Network deployment; dimensioning; mobile broadband; cost-capacity performance; techno-economic analysis; spectrum allocation; wall penetration losses; off-loading; LTE; HSPA I. INTRODUCTION The rapid increase of mobile broadband (MBB) services in combination with flat fee subscriptions has resulted in a decoupling of traffic from operator revenues. Hence, improved cost-efficiency in network deployment and operation is getting more important for operators. One option is to offload data traffic from the macrocell layer to femtocells and hence reduce the need for investments in more costly macrocell networks. Offloading is thus another benefit of femtocells [2][15][18], than improved indoor coverage for voice services that often is discussed for use in homes or small offices. The focus of our analysis is on public access for MBB services. In order to compare the cost-capacity performance of macro and femtocell networks you need to consider other aspects than the often discussed wall attenuation and indoor coverage. In this paper we will analyze deployment for different levels of data usage taking into account existing base stations sites that can be re-used. We will also include the recent development when it comes to cost and spectral efficiency of radio access technologies and also the impact of use of more bandwidth. A number of papers have been published on joint macrocell and femtocell networks, but most papers deal with mechanisms and technical performance rather than deployment strategies and the capacity - cost performance. Throughput for different scenarios are presented in [1][2] and a comparison between open and closed access is presented in [3], data rates for indoor users served by macrocells and femtocells are compared in [4] and capacity and coverage statistics is presented in [5]. Analysis of local and indoor networks has been presented for WiFi hot spots and private networks [7][8][9][10][11]. The focus in these studies is on business models for the local operator, cost and deployment aspects are not covered. Techno economic analysis from a mobile network operator perspective has been performed in a number of large projects like TERA, TONIC and Ecosys, see e.g. [11] for analysis of different deployment strategies and business cases. Analysis of cost structure and deployment strategies for heterogeneous access networks is presented in [12][13]. Femtocell cases are not discussed but a detailed analysis of complementing wide area networks with local WLAN access points is presented in [12]. Analysis of cost structure and business cases for deployment of MBB is presented in [14][15]and analysis of femtocell deployment in [16][17][18]. The conclusions about the profitability differs in these papers but they have the same working assumptions; HSPA type of technology, use of 5 MHz of spectrum and the estimates of cost of radio equipment. If we compare the resulting cost to capacity ratio used in these papers with estimates from 4G contracts late 2009 in the Nordic countries large differences can be identified. Now the cost for radio equipment for a three sector site supporting 20 MHz is well below 10k. The cost capacity ratio is 20-40 times lower than in the referenced papers [14-18]. The objective of this paper is to make a more up to date comparative analysis between macrocell and femtocell deployment taking into account the recent developments in spectrum efficiency, lowered prices of radio equipment and the use of system bandwidth up to 20 MHz. The focus is on public access for MBB services and not on voice services in homes. The paper is outlined as follows; section II describes the analysis approach, the models and assumptions. Section III contains the results and in section VI we discuss issues related to the base station density and the over-provisioning of femtocell capacity. A summary is found in section V.

II. METHODOLOGY, MODELLING AND ASSUMPTIONS In this section we will describe the analysis approach, the assumptions and the scenario that is used in the analysis. We will compare the cost-capacity characteristics of macrocellular and femtocell deployment. The analysis includes a network dimensioning part and a cost structure modeling and analysis part. Additional analysis includes strategies to compensate for wall penetration losses and to guarantee a specified data rate. A. Scenario description We consider construction of a new business park to be built 2010. There are 10 five floor office buildings, each with 1000 persons, in the 1 km 2 area; see Fig 1.The facility owner wants to investigate the options for provisioning of cellular wireless data services in the buildings. We will compare the costs for deployment and operation of networks using outdoor macro base stations with the cost for indoor femtocell solutions. We both consider the cases that all macro base station sites need to be deployed from start and that there are sites that can be re-used. Up to 20 MHz in the 2,6 GHz band will be used for the macro-layer. In the analysis on strategies to compensate for wall penetration losses we include both building a denser 2.6 GHz network as well as the use of the 800 MHz band. B. User demand Our analysis will start with the user demand described by the number of users per area unit and the data usage (GB) per user and month. This is converted to a required capacity per area unit (Mbps/ km 2 ) computed for a number of busy hours. The demand is assumed to be 14,4 GB per user and month. For the dimensioning two levels of wireless usage are considered: 10% and 50 % of the total monthly usage, this corresponds to average user data rates of 20 and 100 kbps respectively during the 8 busy hours.. For operators in Sweden the MBB usage 2009 is 2 5 GB per user and month. In the area in our example we have 10 000 office workers. The assumed 14,4 GByte per month and professional user equals 720 MB per day and 90 MB per busy hour (assuming 8 busy hours per day). The total data demand for the 10000 workers is 2,0 Gbits per second (Gbps). For network dimensioning we need to consider the share that is assumed to be carried over the wireless connections. The low and high level estimates of 10% and 50 % mean 200 Mbps and 1 Gbps for the total wireless usage in the area. TABLE I. CAPACITY FOR A THREE-SECTOR SITE (MBITS PER SECOND) Spectral efficiency (Cell average) Allocated Bandwidth 5 MHz 10 MHz 20 MHz 0,67 bps per Hz 10 Mbps 20 Mbps 40 Mbps 1,67 bps per Hz 25 Mbps 50 Mbps 100 Mbps C. Coverage and capacity of radio access technologies Different number of base stations is needed in order to satisfy the demand depending on the coverage and capacity characteristics of different technologies. We consider two types of radio access technology each with spectral efficiency of 0,67 and 1,67 bps/hz respectively. These numbers can represent current HSPA evolution and future releases of LTE [19][20]. Assuming three-sector sites these values of spectrum efficiency implies a site capacity of 10 Mbps and 25 Mbps for each chunk of 5 MHz bandwidth; see Table I. In our example we need to cover 1 km 2 area using a number of base stations and we need to make sure that the coverage is sufficient for the data rates of interest. We use the analysis in [21] where the radio range is estimated for different frequency bands for a 1 Mbps data service using a LTE type of system and 10 MHz of bandwidth. With the assumptions in [21] on antenna heights, wall propagation losses (20dB) and on antenna diversity the ranges for indoor coverage are 1.5 km at 900 MHz and 0.7km at 2.5 GHz for urban deployment. Since a cell area of 1 km 2 corresponds to a cell radius of 0,57 km our requirements on average user data rates during busy hours (~ 0,10 Mbps) would be met even at the cell borders. Using 20 MHz of spectrum and the technology with higher spectral efficiency the low and high demand levels (0,2 and 1,0 Gbps) can be met by deploying 2 and 10 macro base stations respectively. With the site capacity number in table I we can estimate the number of sites required to satisfy the user demand with different amounts of bandwidth, see Fig 2. Use of the technology with lower spectral efficiency and 5 10 MHz of bandwidth will result in a high number of base stations sites, 50 100, in order to satisfy the demand in the 1 km 2 office area. In our comparison we will use the technology with high spectral efficiency and macro base stations using 20 MHz at 2,6 GHz or 10 MHz at 800 MHz. Analysis of co-channel and adjacent channel operation show that different types of interference will lead to performance degradation [1][22][23][24]. Hence, in this paper we will assume that femtocells are deployed in a separate frequency band Number of base station sites 120 100 80 60 40 Spectral eff = 1,67 (LTE type) Spectral eff = 0,67 (HSPA type) 20 0 5 10 15 20 Used spectrum (MHz) Figure 1. Office buildings in the scenario used in the cost analysis example Figure 2. Number of base station sites as function of allocated bandwidth

For the femtocell deployment we assume access points using 5 MHz of spectrum. In our comparison we will use the value 10 Mbps for the femtocell capacity. However, with the levels of user demand assumed in this study (< 0,1Mbps) the femtocell capacity will not be the limiting factor, see also discussion in section V. D. Network dimensioning The macro base station network is dimensioned to meet the user demand in terms of average busy hour data rate over the whole area. Users are allocated to base stations assuming sharing of the offered capacity, e.g. 10 users can get on average 0,5Mbps when sharing a 5 Mbps base station assuming best effort type of usage. For the dimensioning of the femtocell networks we will use two other approaches: User oriented; a number of users (2 or 8) are allocated to each access point (no matter the demand) Coverage oriented; a number of access points (8 or 16) are allocated for each floor in the building However, the average busy hour data rate per user does not reflect any requirements of guaranteed availability or data rate. Hence, in the discussion section we will provide examples of dimensioning taking into account the data rate guarantees resulting in the probability for a user to get a specified data rate provided a specified of capacity of the base station. E. Impact of propagation losses 1) Wall attenuation for macrocell deployment To estimate the effect of the total wall attenuation we use the Okumura-Hata propagation model for an urban macrocell [24]. For a certain data rate a maximum propagation loss can be calculated. Assuming the maximum cell range to be R (km), the total wall attenuation = W (db), K macro a constant and a (h) a correction for the height h of the mobile antenna, then the propagation loss L can then be expressed as L = K macro + 35,2 log(r) + W a (h) (1) If we assume an antenna height of 30 m for the macro base station and the mobile receiver to be 1,5 m above ground then for frequencies 800-900 MHz the constant K macro = 125 126 and for frequencies 2,1 2,6 GHz K macro = 136 138. While h is about 1,5 m for ground floor it may be about 10 m for the third floor. In the latter case a (h) is about 5 db. 2) Propagation and wall attenuation for femtocells Inside a room the criteria for free space propagation are met implying the formula for the propagation loss L db. L = K femto + 20 log D (2) (K femto is a constant; D is the distance femtocell terminal) K femto = 32,4 + 20 log (f) (K femto = 98,8 for f = 2,1 GHz) (3) Outside the room we have to take into account wall attenuation W and floor attenuation F. To the free space loss above we have to add the wall and floor losses which are proportional to the number of walls n and floors m the signals have to penetrate. The equation now reads: L = 98,8 + 20 log D + n W + m F (4) At a distance D of 50 m L = 72 + nw + mf This figure is much smaller (about 60-70 db) than the propagation loss from the macrocell when the outer wall attenuation (15 25 db) is taken into account. This implies that very low power can be used for the femtocell. The attenuation between two adjacent rooms is normally only a few db while the attenuation between floors is much higher (up to 20 db or more) of construction reasons. This has advantages from interference point of view between the femtocells. The risk of co-channel interference is reduced in particular between femtocells on different floors. F. Cost structure modelling and analysis The cost structure analysis is in the form of a comparison of capacity expansion using only femtocells or macro base station solutions satisfying the same user demand. For the modeling of capacity demand and the cost structure break down we use the methodology proposed and used by Johansson [12]. We will take into account both capital expenditure (CAPEX) and operational expenditure (OPEX) for the two types of deployment. The main elements in the cost structure model are: Investments in radio equipment, base station sites and transmission, installation and deployment costs, site leases, costs for leased lines and costs for operation & maintenance. We have assumed that the cost for deploying a new macro base station site in the urban area is 100 k including transmission. Cost for upgrading an existing site is estimated to 10 k. The cost for radio equipment supporting three sectors and 5 20 MHz is estimated to be 10 k. Estimates of the OPEX components are shown in table II. In the analysis we will use the number 30 k per year for a new site and 10 k per year when an existing site is re-used. For the femtocell solutions we use data from a WLAN large scale deployment project running over a couple of years and with several hundred access points 1. On average the deployment of one access point is 1 000 with roughly equal shares (20-30%) for the following cost components: Access point equipment, planning & installation, transmission and the share of the AP controller and management system. For the femtocells we estimate the annual operational cost to be 500 per access point. This corresponds to a broadband connection with quite a high yearly price. For both macro and femtocell networks transmission costs are included assuming that a backbone is already build out. TABLE II. ESTIMATES OF OPEX FOR MACRO BASE STATION SITES OPEX component Annual cost Comment Site lease 5 10 k Downtown/City area Leased lines 12 k 1-2 k per E1 O&M 5 10 k 5 10 % of total CAPEX Power 3 5 k TOTAL 25 37 k 1 Interview with Per Lindgren, David Lundberg, Uppsala University

III. RESULTS A. Base case With the assumed capacity and CAPEX figures for the macrocell and femtocell solutions it is straightforward to estimate the number of base station sites and femtocell access points using the different approaches outlined in section II D. These numbers together with estimated CAPEX and deployed capacity is presented in tables III and IV. Our initial assumption is that new macro base station sites need to be deployed. For the low demand level the CAPEX for femtocell and macrocell networks are equal. With the coverage approach the femtocell solutions are cheaper for the high demand level. Using the user oriented approach with 4 or 8 users per access point the femtocell network is more expensive but the cost is in the same order of magnitude as the macrocell network. However, this situation is drastically changed if i) existing macro sites can be re-used or if ii) the wall penetration is so large that it need to be compensated for in some way, see next subsection. The case with re-use of base stations sites is included in table IV, i.e. no site costs are included which results in lower costs than for any of the femtocell solutions. B. Case with compensation of wall penetration losses We will do a sensitivity analysis for the case where the wall penetration losses exceed the assumed 20dB used in the base case. The outer wall attenuation varies very much and can be up to 25 db or more for concrete or brick walls in urban areas. The walls between rooms in a building normally have a few db of attenuation. So inside a building the total attenuation is the sum of the outer wall attenuation and a number of inner walls attenuation and possibly also from floors. Let assume that we need to compensate for another 12 db of attenuation. TABLE III. Coverage approach INVESTMENTS AND DEPLOYED FEMTOCELL CAPACITY Number of femtocells CAPEX Total capacity 4 femtocells per floor 200 0,20 M 2000 Mbps 8 femtocells per floor 400 0,40 M 4000 Mbps User oriented approach TABLE IV. 8 users per femtocell 1250 1,25 M 12 500 Mbps 4 users per femtocell 2500 2,50 M 25 000 Mbps Deploying new sites Reusing existing sites INVESTMENTS AND DEPLOYED MACRCELL CAPACITY Number of sites CAPEX Total capacity Low level demand 2 0,24 M 200 Mbps High level demand 10 1,20 M 1000 Mbps Low level demand 2 0,04 M 200 Mbps High level demand 10 0,20 M 1000 Mbps TABLE V. CHANGE OF BASE STATION DENSITY (N) REQUIRED TO COMPENSATE FOR ADDITIONAL WALL ATTENUATION (W) Wall Attenuation (db) N 0 1 5 2 10 3,7 15 7 20 14 We will look into two compensation strategies Deployment using the 800 MHz band To build a denser network in the 2.6 GHz band The difference in path loss between operation in the 800 and the 2.6 GHz band is around 12 db. Hence, the same number of sites as in the 2.6 GHz base case could be used, i.e. 2 and 10 for the different demand levels. According to our assumptions only 10 MHz is available in the 800 MHz band and using table I we can conclude that we need 4 and 20 base stations sites for the low and high demand levels respectively. If no 800 MHz band is available the operator needs to decrease the cell size for the 2.6 GHz network. By using the formula (1) for different values of the wall penetration loss W the corresponding values for cell ranges R that compensates for the loss W can be calculated. This can be used to calculate the number N of base stations to cover the 1 square km area. In table V we can see that W ~ 12 db corresponds to N = 5. In the summary in table VI we can see that all of the strategies that lead to deployment of a large number of new sites are very costly. Use of the 800 MHz band is comparable to femtocell deployment cost for the low demand level but higher for the high demand level. However, re-use of existing sites is very cost-efficient even when many sites need to be equipped with new radio transceivers. TABLE VI. INVESTMENTS AND DEPLOYED MACROCELL CAPACITY IN ORDER TO COMPENSATE FOR WALL PENETRATION LOSSES Deploying new 800 MHz sites Reusing sites for 800 MHz Number of sites CAPEX Total capacity Low level demand 4 0,48 M 200 Mbps High level demand 20 2,40 M 1000 Mbps Low level demand 4 0,08 M 200 Mbps High level demand 20 0,40 M 1000 Mbps 5 times more new 2.6 GHz sites Low level demand 10 1,20 M 1000 Mbps High level demand 50 6,00 M 5000 Mbps 5 times more re-use 2.6 GHz sites Low level demand 10 0,20 M 1000 Mbps High level demand 50 1,00 M 5000 Mbps

TABLE VII. CAPEX,OPEX AND NET PRESENT VALUE (HIGH DEMAND) TABLE VIII. EXAMPLES OF BASE STATION DENSITIES (URBAN AREAS IN SWEDEN ) Deployment type CAPEX (M ) OPEX (M ) NPV (M ) Femtocells "8 AP per floor" 0,4 0,2 1,31 " 8 users per AP" 1,5 0,625 4,34 Macro 2,6 GHz Base case - new sites 1,2 0,3 2,77 Base case - re-used sites 0,2 0,1 0,65 5 times denser - new sites 6 1,5 12,82 5 times denser - re-used sites 1 0,5 3,27 Macro 800 MHz New sites 2,4 0,6 5,13 Re-used sites 0,4 0,2 1,31 IV. COST ANALYSIS In order to fully compare different deployment options we also need to include the operational expenditure (OPEX) in the analysis. Table VII presents the resulting CAPEX, OPEX and Net Present Value (NPV) for different types of deployment. The NPV analysis is done for 5 years assuming that all investments are made year 1, with a discount rate of 5 % and assuming that the OPEX is increasing 10% each year. From a cost perspective the choice between solutions based on femtocells or macrocells is not primarily about the cost of the radio equipment. The key issue is if new macro base station sites need to be deployed or not. This drives the CAPEX but also result in higher OPEX compared to the case where sites can be re-used and the OPEX can be shared between the existing (voice) services and the mobile broadband services. It is also shown that a dense deployment of femtocells is less cost efficient. Even if the femtocells were for free, the installation, cabling and operation results in high overall costs due to the large number of femtocells. We believe that the most important characteristic of the femtocell deployment is that it provides usable resources. Compared to the macrocellular case there is no need to deploy additional capacity in order to compensate for propagation loss, see also section IV B. Name and type of area Total density of sites Typical densities for operators Residential area in Uppsala ~6 per km 2 1 3 per km 2 Residential area Akalla ~ 14 per km 2 3 5 per km 2 Central part of Uppsala ~ 20 per km 2 3 8 per km 2 Industry area Kista ~ 50 per km 2 7 20 per km 2 Central part of Stockholm ~ 130 per km 2 20 40 per km 2 a. Numbers derived from PTS Transmitter map web page, December 2009 V. DISCUSSION A. Density of macro base station sites In the analysis of our example cases we have seen both low numbers of base station densities, 2 4 per km 2, as well as large numbers, 20 50 per km 2. The latter figure is for the high demand level in combination with denser deployment in order to compensate for additional wall penetration losses. Are these numbers realistic compared to real deployment scenarios? A small survey of the locations of existing base station sites in the Stockholm area results in the base station densities shown in table VIII. These results indicate high densities in industry and downtown areas. In total 50 100 sites per km 2 and 10 40 sites per km 2 for each operator. Hence, in areas with existing sites the strategy to meet the demand for mobile broadband services should be based on site re-use and deployment of macro base stations using as much bandwidth as possible in the bands available for the operator. However, in areas without any existing infrastructure the situation is different. The high density of base station sites, see also Figures 3 and 4, implies small cells with cell radiuses in the order of 100 200 m. Hence, the link budget would be good enough in order to ensure reasonable high data rates even at indoor locations at the cell borders. Figure 3. Base station sites in Kista area (circles = GSM, squares = 3G) Figure 4. Base station sites in Stockholm (circles = GSM, squares = 3G)

TABLE IX. NUMBER OF SERVED FEMTOCELL USERS USING THE TWO APPROACHES AVERAGE THROUGHOUT AND GUARAANTEED BIT RATE Cell capacity No served users with average throughput 0,10 Mbps No users with 95 % probability of being served with 1,0 Mbps 2 Mbps 20 4 3 Mbps 30 9 7 Mbps 70 37 B. About over-provisioning of femtocell capacity In all the femtocell deployment examples above the resulting capacity is well above the user demand levels. The reason for this is the used dimensioning approach based on limited number of users per femtocell base station or the number of base stations per floor. This over-provisioning of capacity can be exploited in two ways: i) to provide guaranteed data rates, ii) to offer a future proof solution for higher demand levels. The real office traffic may be of a different nature with a number of transmissions of varying length at a higher data rate suitable for the actual service. Many applications require 1 Mbps or more. Using tele-traffic theory (and the Erlang B loss formula.) we can estimate the probability of the users getting service when guaranteeing a certain data rate (e.g. 1 Mbps) for various cell capacities. For the case of an average data rate of 0,1 Mbps the usage would be 10% of the time corresponding to a traffic load of 0,10 E (Erlang). In table IX we show the difference in number of served users between the two approaches average throughout and guaranteed bit rate. It illustrates how the non used femtocell capacity can be exploited when only few users are connected to a femtocell base station. Femtocells that can serve more users (16 32) have recently been announced. VI. CONCLUSIONS We have compared the cost-capacity performance for macrocellular and femtocell networks. For the low demand levels in the investigated scenario the macro solutions using 20 MHz of access technology with high spectral efficiency results in lowest cost. The femtocell solution is less cost efficient due to the need to deploy a large number of access points in order to ensure coverage. For the assumed high demand level the solutions are more or less similar, the main cost drivers being the build out of base station sites and the deployment of many femtocells. Due to the over provisioning of capacity a dense femtocell network can satisfy very high demand levels. The analysis indicates that the most important aspect when comparing femto and macrocell solutions is if new macro sites need to be deployed or not. When a dense macro network with new sites is needed the femtocell solution has much lower cost. However, if existing macro sites can be re-used the macro solution shows is much more cost efficient. The re-use of sites have a large impact also when a denser macro network is deployed in order to compensate for wall attenuation. An option to investigate further is the combination of a macro layer solution, using existing sites and as much available spectrum as possible, with a supporting femtocell network. REFERENCES [1] D. López-Pérez et al, OFDMA Femtocells: A Roadmap on Interference Avoidance. IEEE Communications Magazine, September 2009. [2] D. Calin, H. Claussen, H. Uzunalioglu, On femto deployment architectures and macrocell offloading benefits in joint macro-femto deployments, IEEE Communications Magazine, Jan. 2010 [3] G. De la Roche, A. Valcarce, D. López-Pérez, J Zhang, Access control mechanisms for femtocell,ieee Communications Magazine, Jan. 2010. [4] J. Weitzen, T Grosch, Comparing coverage quality for femtocell and macrocell broadband data services, IEEE Communications Magazine, Jan. 2010. [5] M. Yavuz et al, Interference management and performance analysis of UMTS/HSPA+femtocells,IEEE Communications Magazine, Jan. 2010. [6] Lindner T et al, 2004, Exploitation of Public and Private WiFi Coverage for New Business Models, IFIP Conference on E-Commerce, E-Business, and E-Government (Springer) [7] Bar F, Park N, Municipal WiFi networks: The goals, practices and policy implications of the US case, Communications & Strategies No 61, 2006, [8] Markendahl J, Mäkitalo Ö, Business model analysis for local access provisioning, The 18th Int. Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC707); 2007 [9] Cox J, Municipal WiFi as large scale for profit projects falter innovative new models emerge, Network World: April 21, 2008 [10] Middleton C, Potter A, 2008, Is it good to share? A case study of the FON and Meraki approaches to broadband provision, ITS Biennal Conference, Montreal, June 2008 [11] Loizillon F et al., "Final results on seamless mobile IP service provision economics", IST-2000-25172 TONIC Deliverable number 11, Oct 2002. [12] Johansson K, Cost Effective Deployment Strategies for Heterogeneous Wireless Networks, PhD Dissertation, Royal Institute of Technology, Stockholm, 2007 [13] Johansson K, et al, "Relation between base station characteristics and cost structure in cellular networks", 15 th Int symposium on Personal, Indoor and Mobile Radio Communications (PIMRC 04), 2004. [14] Blennerud G, Don t worry Mobile Broadband is profitable, Ericsson Business review, No 2, 2009 [15] Mölleryd et al Analysis of operator options to reduce the impact of the revenue gap caused by flat rate mobile broadband subscriptions, 8 th Conf. on Telecom, Media & Internet Tele-Economics, June 2009 [16] Claussen H, Ho L, Samuel L, 2007, Financial Analysis of a Pico- Cellular Home Network Deployment, Proceedings ICC 2007 [17] Markendahl J, Mäkitalo Ö, Werding J, Analysis of Cost Structure and Business Model options for Wireless Access Provisioning using Femtocell solutions ; 19 th European Regional ITS Conference, 2008 [18] Femtocell Business case White Papers; June 2009 and February 2010, www.femtoforum.org [19] Bergman J et al, Continued HSPA Evolution of mobile broadband, Ericsson Review, No 1, 2009 [20] Parkvall S, Astely D, The Evolution of LTE towards IMT-Advanced Journal of Communications, vol. 4, No. 3, 2009 [21] Holma H. & Toskala A. (ed), WCDMA for UMTS - HSPA evolution and LTE, John Wiley & Sons, 4th edition., 2007 [22] Claussen H, Performance of Macro- and Co-Channel Femtocells in a Hierarchical Cell Structure, IEEE 18th Int. Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC 07), 2007 [23] Ericsson, ``Downlink co-existence between macro cells and adjacent channel Home NodeBs.", 3GPP TSG-RAN WG4, R4-080939, Meeting \# 47, Kansas City, USA, 5th - 9th May, 2008. [24] Espino J, Markendahl J, Analysis of Macro - Femtocell Interference and Implications for spectrum allocation, The 20th Int. Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC 09); 2009 [25] Hata M, 1980, Empirical Formula for Propagation Loss in Land Mobile Radio Services, IIIE Tr. Vehicular Technology, Vol. VT-29, No.3, 1980