Identifying Some Techno-Economic Criteria in PLC/BPL Applications - - PDF document

Identifying Some Techno-Economic Criteria in PLC/BPL Applications and Commercialization Paul A Brown

White Box Associates 30 Applerigg Kendal, Cumbria, LA9 6EA, UK E-mail: pab@whiteboxassociates.co.uk

ABSTRACT The development

• f

first-pass simplified business models for power line communication (PLC) systems is necessary in order to provide an early focus on the key technical and financial criteria by which investment and the return on investment (ROI) might be considered. For example, PC based spreadsheet modeling provides a useful mathematical tool for identifying sensitivities in the broadband power line (BPL) business cases by enabling rapid comparisons between electrical distribution network (EDN) technical parameters, BPL equipment specifications, system features and customer expectations. It is important for both developers and investors to understand the technical challenges and business opportunities

• ffered by PLC.

1. INTRODUCTION

PC based spreadsheet modeling and graphical analysis provide a first-pass means of identifying and comparing the sensitivities of both technical and financial input data such as EDN topography, channel characteristics, PLC system cost, capability, capacity, quality of service (QoS) and potential penetration rates in order to support an acceptable ROI. In order to develop PLC systems there are a number of fundamental requirements:

• A power line infrastructure i.e. EDN
• Investor(s) in PLC R&D
• PLC developer(s) with the necessary

expertise The power line is a pervasive element and is therefore not a problem. Investors need to feel confident that there is a viable and potentially long term business opportunity in PLC. PLC developers need engineering expertise, investment and a detailed knowledge of EDNs as a means of providing a sustainable transmission medium for telecommunication services. Investors do not necessarily possess engineering expertise and therefore they require the developer’s skill and knowledge. Similarly the developers need financing in order to sustain PLC development. So what kind of first-pass information might be useful to investors? If we assume that PLC / BPL is now at a stage of potential mass-deployment and R&D continues to enhance PLC systems development then, most

• f all, we require to scope out the potential scale
• f the ROI.

For example, investors need to know such things as:

• Is the PLC business case sensitive to

equipment costs such as customer premises equipment (CPE), repeaters, coupling devices etc?

• Is BPL system design sensitive to

service transmission speeds e.g. 256, 512 kb/s etc?

• Does EMC regulation impact PLC /

BPL system design?

• Is voice over Internet Protocol (VoIP)

achievable over PLC systems?

• What might be the anticipated churn

rate?

• What are the potential PLC / BPL
• perational and marketing overheads?

Whilst potential customers might, for example, focus on the following issues:

• Installation fee
• Monthly, quarterly or annual service

charges

• QoS parameters

• 2. CONSIDER BPL NETWORK TOPOLOGY

In a first-pass BPL business plan it will be necessary to make some assumptions about the network topology. These assumptions are needed in order to consider specific deployment scenarios whilst still being general enough to be applied to a wide variety of network topologies. The PLC network is of necessity a hierarchy of

• networks. This global perspective of electricity

and telecommunication transmission and distribution (T&D) infrastructure is illustrated in Figure 1 and is shown as a hierarchy of electricity distribution (wired) and telecommunication transmission (fiber optic) networks.

• Fig. 1: PLC – Hierarchy of Networks

Figure 2 illustrates a section of typical UK low voltage (LV) underground (UG) EDN interconnecting residential customer premises to the local LV substation transformer.

• Fig. 2: Section of a UK LV EDN Diagram

Figure 3 shows an overall schematic view of the network infrastructure including the integration

• f domestic (residential) customers, small

businesses and corporate telecommunication

• networks. It also illustrates how these networks

further interconnect utilizing PLC overlays on LV EDN, medium voltage (MV) EDN and incorporates fiber optic backhaul. These network combinations include metropolitan area networks (MANs) and the interfaces to Internet Service Providers (ISPs). This model is typical

• f most all of the major European BPL

deployments to date.

• Fig. 3: Fiber Optic MV and LV Networks
• 3. DATA INPUT VERSUS OUTPUT

If we consider, for example, EMC issues which affect PLC system operation then we might choose to examine a range of relevant input parameters such as: the PLC equipment launch power spectral density (PSD), the average attenuation of the channel, the channel noise floor, and the minimum signal to noise ratio (SNR) required for adequate operation of the PLC equipment. Research in Europe [1] has shown that with a nominal channel attenuation (ATT) of 0.4 dB/m (i.e. below 10 MHz), a flat noise floor PSD of -120 dBm/Hz and a minimum SNR of 15 dB the distance (D) covered by a single customer premises modem, or repeater, might be estimated as follows: D max = (PSD tx – PSD noise – SNR min) = ATT dB/m = (- 50- (-120) – 15)m = 137.5 m (1) 0.4 If we now model the average LV transformer to building distance (e.g. by reference data gathered from the respective EDN) and assuming a best- fit Gaussian distribution, with an average distance of, for example, 100m and a standard deviation of say 50m. We might then use this statistical distribution to estimate the cumulative statistical distribution of the distances between the buildings and the LV EDN transformer. From this initial output data we might choose to

MV EDNs LV transformer - substation LV EDNs Fiber Optic MANs

LV S/Stn PLC Repeater Backbone Network LV S/Stn

MVEDN LVEDN

Corporate Intranet PLC Concentrator MAN SDH Fibre Core Network Management ISP Gateway

further graph ‘Distance to the LV Transformer’ against ‘Percentage of Buildings (customer premises) served by a transformer. The graphical results obtained for such an example are illustrated in Figure 4.

• Fig. 4: Reach Vs Percentage of Buildings

As can be seen from Figure 4 approximately 80% of buildings are below the 137.5m threshold and might therefore be overlayed with BPL without the need for a repeater. Now from Equation 1 and interpolating for a BPL launch PSD of -50dBm/Hz we can reach 137.5m, at maximum, from the substation transformer without the need for repeaters. Therefore we already have some cost inference from our modeling i.e. there will be additional cost in providing and installing a repeater if it is required to extend the reach to pass a further 20% of the customers premises and still comply with EMC requirements (i.e. launch PSD does not exceed -50dBm/Hz). So now we have a simple example of EDN input data (i.e. parameters such as cable attenuation, launch PSD...etc) being formulated (Eq. 1) and the resulting data output in graphical form. Equation 1 requires only simple mathematical

• perations and can easily be applied in

spreadsheet form (e.g. Excel) and we can therefore easily modify the input data in order to test sensitivities.

• 4. EMC – CUMULATIVE EFFECTS

Practical EMC measurements of both the input and output parameters of PLC networks (i.e. launch PSD and radiated emission levels) and their subsequent comparison with calculated values are of considerable importance in PLC system development [2]. Therefore we might consider some further aspects of using a diagrammatic appoach to depict physical data

• sources. First by examining a particular network

element in isolation and then when integrated with a number of similar elements such as might be found in practice. If we then examine the PLC network component in isolation and then integrated with other similar components and by applying a variety of, potentially, relevant mathematical formulas we might make comparisons between calculated and measured input / output parameters. This should enable validation of the most applicable formula whose calculated results align most accurately with the practical results obtained by measurement. For example, if we assume a single PLC cell as a first-pass model (i.e. a single element), as shown in Figure 5, then integrate this single cell with a number of similar cells, as we might expect in practice, we might then deduce the cumulative effect of a number of such cells on the noise floor at a given distance [3].

• Fig. 5: Single PLC Substation Cell

Figure 6 shows the integration of a number of such PLC cells around a notional ‘quiet’ radio site where the noise floor is, approximately, -15 dBm/Hz measured in a 9kHz bandwidth.

• Fig. 6: Multiple PLC Substation Cells

Now we can visualize the situation we might choose to compare the possible propagation modes, at or near the Earth’s surface, which

nMbit/s (multiplexed)

Electricity & Telecomms Electricity PLT Basestation

The PLC S/Stn Cell

UK typical UG Distributor 250m 150 Customer Premises per S/Stn

might in turn help to describe the cumulative effects of such a PLC deployment on the noise floor. From above, Equation 2 provides a ‘Free Space’ propagation model, Equation 3 provides an ‘Over a Plane Earth’ model and Equation 4 provides an ‘Egli’ propagation model. So now we might utilize a spreadsheet approach to graph the input and output data for these three

• equations. This is illustated in Figure 7.
• Fig. 7: Scatter Plots and Trend Lines

Once we have established the most closely aligned theoretical (calculated) and measured data sets (i.e.

• verlaid

scatter plots

• f

measurement results) from our three theoretical propagation models, as shown in Figure 7, we might then choose to extend our consideration of the cumulative impact of additional PLC cells on the noise floor. We might choose to output this data, for example, in a tabular format as illustrated in Figure 8.

• Fig. 8: Incremental Increases in Noise Floor

So we have now illustrated how we might generate a data capture rationale and how we might manipulate, model and depict results which focus

• n

some

• f

the technical considerations for BPL networks and systems (i.e. EMC criteria) and which ultimately impact system economics.

• 5. CONSIDER COMPETITION BPL-ADSL

BPL deployments may take place in locations where there is competition in the broadband

• market. This will impact potential BPL service
• fferings and price, but if the competition (e.g.

DSL, CATV etc) is not considered in sufficient detail BPL service penetration may be negligible. For example, we might relate BPL Service Penetration (SP) to Service Cost (SC), including installation cost with the logarithm of the Data Transmission Rate (DTR) as follows: SP = (60-SC)+10x(log2(DTR)-9)% (5) 100 Again we have derived a first-pass relatively simple formula, which might easily be applied with output data presented in a spreadsheet format with the results tabulated or in graphical form.

• 6. BANDWIDTH-TRAFFIC KEY FACTORS

Bandwidth is a key factor for any transmission system and PLC / BPL is no exception. If, for example, we introduce repeaters into a system we might increase reach without increasing the

12th October 2001 - Measured E Field Strength regression @ 7 MHz, +10dBm injected signal power into CPCU, Dipole, EGLI Propagation Model (Dipole HT=6m HR=2m, and CPCU HT=1.5m HR=2m), and Free Space Model.

• 25
• 20
• 15
• 10
• 5

5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 50 100 150 200 250 300 350 400 450 500

Distance (m) F ie ld S tre n g th (d B u V /m )

Free Space Theoretical EGLI Dipole Theoretical EGLI CPCU Theoretical Dipole Injection Dipole Trendline CPCU Injection CPCU Trendline

launch PSD, but of necessity we will incur penalties in the frequency and/or time domains and also, to some extent, in throughput delay. However, bandwidth partitioning and bandwidth allocation are both service and application

• dependant. Internet services tend to have bursty

transmission characteristics, similar to those of Intranet local area network (LAN) services where there are shared resources which contend for the communication channel bandwidth [4]. So we first need to examine the probability of a number

• f

users wishing to share the communication channel simultaneously. Assume ‘n’ customers each with a probability ‘p’ of being connected at any given time. Then the probability of ‘k’ customers being connected at the same time is given by: (6) Then if we assume 20 customers (n=20), each with a 10% probability

• f

simultaneous connection we get the graphical data output from the spreadsheet as shown in Figure 9. As can be seen there is a 30% probability that 2 customers will require to be connected simultaneously, a 19% probability for 3 customers, a 9% probability for 4 customers etc. Clearly for 7 or more customers there is little probability of a simultaneous requirement, but a first-pass design figure will be required, so initially we might, for example, wish to examine a requirement for 5 customers to be connected simultaneously.

• Fig. 9: Prob. Of Simultaneous Connection

Then by distributing bandwidth to cater for 5 simultaneous customers we have the probability distribution shown in Figure 10.

N.B. In 95% of cases less than 5 customers connected

• Fig. 10: Bandwidth Provision for 5 Customers

Bandwidth distribution will be a critical issue and will directly impact the QoS. It is therefore a key techno-competitive issue for BPL which can be modeled, in some detail, by spreadsheet analysis.

• 7. COMMERCIAL – OPERATIONAL ISSUES

We have overviewed some of the technical, service-application and competitive issues which impact PLC / BPL roll-out [5] and observed some relationships from spreadsheet analysis, which ultimately contribute to describing the

• verall system deployment cost. However, this

‘inside-out’ perspective might well be complimented by examining the costing processes from the ‘outside-in’. For example, the major economic issue will, most probably, be broken down into expenditure, revenue and

• profit. Clearly the process must commence with

investment in R&D, so we have this as an expense from day 1. Then we might proceed to technical trials, with additional expenditure and investment and finally to roll-out and yet more

• expenditure. But after roll-out we have revenue

streams and return on the initial investment, however by this time many years may have

• passed. So we might expect to see economic

predictions, say over an initial 5 year period, as shown in Figure 11.

• Fig. 11: BPL Expenditure, Revenue & Profit

• 8. DEVELOPING ECONOMIC FORECASTS

From the example in Figure 11, we now need to correlate expenditure and revenue in order to determine profit and loss over the same 5 year period and this is summarized in Figure 12. From Figure 12 we can see that in year 1 we have a loss, similarly in year 2, but in the following years 3, 4, and 5 we have profit, increasing each year. So we have for this example a forecast, as we might expect, of dominant expenditure in the first 2 years followed by 3 years in which revenues and consequently profit dominates.

• Fig. 12: BPL Yearly Profit to Investment Ratio
• 9. MICRO - MACRO APPLICATIONS

It can be seen with reference to Figures 4, 9 and 10 that there is a micro application of spreadsheet modeling where we consider only some relatively small component or part of the data set in relative isolation. We can set this micro perspective against the data in Figures 11 and 12 where we take a macro view of financial

• issues. Each is important and the micro analysis

must, at some point be included and will cumulatively impact the macro perspective. Then we might better understand and quantify the business opportunities for PLC / BPL.

• 10. RATE OF RETURN ON INVESTMENT

Whether an R&D project is funded by government by private investors or other consortia a key element will be for the realistic and accurate translation of the micro techno- economic issues into a macro project and/or business proposal. The proposal will required to be supported by a robust and sufficiently detailed business plan. When such projects contest for financial backing there is a basic requirement to project an ROI as accurately as possible and also to provide a means to test the sensitivities to which the plan may be subjected during the lifetime of the project or business opportunity.

• 11. CONCLUSIONS

As

• ur

previous examples have shown spreadsheet analysis and particularly graphical analysis is a useful tool to observe the dependencies, interactivity and trends in a wide variety of data. Therefore it is concluded that such techniques have an important role to play in the development, assessment and comparison of PLC / BPL investment data, particularly in regard to the embedded techno-economic issues. ACKNOWLEDGEMENTS The author wishes to thank the European Commission and the IST Project 6POWER Consortium for permission to reproduce a number of the figures in this paper. Also, the author wishes to thank White Box Solutions for permission to reproduce a number

• f the figures in this paper.

REFERENCES [1] C. Gomez and J. Palet, ‘Business Plan Update’, D1.2B, 6POWER, European Commission, IST Project No. IST-200137613, November, 2004. [2] F. Issa, D Chaffanjon and A. Pacaud, ‘Outdoor Radiated Emission Associated With Power Line Communication Systems,’, Proc. IEEE International Symposium

• n

EMC, Montreal, Canada, 2001, pp. 521-526. [3] D. Fenton and P. A. Brown, ‘Modeling Cumulative High Frequency Radiated Interference from Power Line Communication Systems’, Proc. 6th ISPLC, March 2002, pp. 83- 87. [4] H. Hrasnica and R. Lehnert, ‘Investigation of MAC Protocols for Broadband PLC Networks Under Realistic Traffic Conditions’, Proc. 7th ISPLC, March 2003, pp. 173-178. [5] A. S. Escalona, ‘PLC Commercial Deployment of Iberdrola’, Proc. 8th ISPLC, March 2004, pp. 171-174.