Techno-economic price-worthiness of mobile networks: a hedonic heuristic perspective

Abstract

The present study is conceived to develop a model based on hedonic heuristics to deduce a performance index of service products offered by modern mobile telecommunication operators. The index so obtained includes the associated “price-worthiness” (value-for-money) aspects of the services; and, it is adopted to compare and evaluate the relative performance of competitive mobile networks deployed in a service area, supporting App-intense, smart, mobile-devices concomitant to traditional feature phones. Relevant operational details (such as technology-centric mobile-speed parameter of the services rendered) and the associated economic considerations are fused judiciously with hedonic perspectives of the users, in order to infer an overall performance metric for the mobile networks in question. Data availed from typical service areas in the U.S. relevant to specific mobile networks are gathered and a comparison of services rendered is made using the proposed measure. The measure of hedonic considerations specified here as the hedonic pricing index (HPI) leads to corresponding results on price-worthiness of underlying service products versus the techno-economic features; and, relevant aspects of certain incumbent networks in a selected service area in the U.S. are deduced, compared and discussed. Foreseeable limitations of HPI in ascertaining the mobile service performance are identified and discussed.

Appendices

Appendix 1

1.1 Estimating mobile speed index (MSI): An outline

As described in [23], the MSI can be estimated via a logistic regression of a set of variables {z_1, z_{2, …} z_i….,} denoting techno-centric parameters (listed below) with reference to a network and services of interest. Hence, the following list of such technology-specified details is obtained pertinent to a service area for a service period of interest as regards to specific mobile service providers/operators rendering the service type being assessed.

The techno-centric parameters needed to computer the MSI are: Average and maximum values of download link speed in Mbps (DLS) (z₁), average and maximum values of upload link speed in Mbps (ULS) (z₂), average web download speed in Mbps (z₃), probability of success of user-data protocol (UDP) stream transport (z₄), probability of success of HTTP download transport (z₅), probability of 500 kbps successful peer-to-peer streaming (z₆), proportion of unspecified bit rate (UBR) downloads (z₇), proportion of web-page completion (z₈), proportion of successful transit of 3G in 3G-to-4G transitions (z₉), (mean) time-to-first-byte (TFB) (z₁₀) and consistency parameter on repeatable results of performance (z₅).

Algorithm – a summary: Suppose Z is equal to (a₁z₁ + … + a_iz_i + …) with, {z_1, z_{2, …,} z_i.…,} denoting the contributory set of predictive regressors of techno-centric parameters listed above (where each dimensioned entity is taken in a normalized form, as necessary); and, the set {a₁, a₂, …, a_i, …,} are regression coefficients properly assigned to ‘weigh’ the predictive variables. That is, the regression coefficient prescribes a prorated influence on how each explanatory variable contributes to the probability of the regressed outcome. (The larger the regression coefficient, the higher is its influence on the probability of the outcome. The normalization indicated above implicitly provides the necessary weighting).

Next, a logistic regression of Z via the logistic function f(Z) = 1/[1+ exp.(− Z)] = [1/2 + 1/2 × tanh(Z/2)] is performed and the evaluated value of f(Z) yields the required mobile speed index (MSI) as detailed in [23]. A summary of computed exemplary results on the MSI values (expressed as normalized fractions) of incumbent mobile carriers (Site: Boston, MA: USA – 2010 through 2013) is depicted in Table 7 of Appendix 3.

Appendix 2

1.1 Estimating the relative techno-economic performance index (RTPI): An outline

The RTPI evaluation is described in [23 with reference to mobile services rendered in a service area during a specified service period via a set of service types by incumbent network operators. Using the associated techno-centric parameters, the MSI values are first ascertained; and, the RTPI is determined using the following predictive estimators: (i) Techno-centric parameters specified via the estimated MSI values and (ii) normalized details on economics-related factors specified by: (a) Relative status of population (POP) in the service area with reference to the start-year value; (b) relative volume of sales of the service (PPS) in question in the service area, (POP × relative sales in that year); (c) relative per capita income (PCI) in the service area; and (d) a set of normalized productivity-specific economic parameters as described in [23}: {X}_{1, 2, …, 8.}

Algorithm - summary: RTPI is evaluated via logistic regression of the associated techno-economic parameters as follows: First, suppose (ERI)|_year is considered as a factor that implicates the mobile service evolution in the service area due to economic reasons. It would then correspond to a logistically-regressed value of ξ|_year defined as: [Sum of all the normalized economics-related parameters] = {[(POP) + (PPS) + (PCI) + (ΣX_{1,2, …,8})]}|_year. Hence, the logistic regression of ξ, namely, p(ξ) is determined using a logistic regression function. Explicitly, p(ξ) = 1/[1+ exp.(− ξ)] = (1/2) + (1/2) × tanh(ξ/2) ≡ [ERI|_year = (0 to 1)

Next, the (RTPI)|_year: η|_year that decides the overall mobile service performance in the service area inclusive of both techno-specific MSI and economics-based ERI. As such, in [23], it is hypothesized that the hybrid of MSI and ERI values would correspond to a statistical mixture value of:{[MSI]_year} and {[ERI]_year} specified as: η = {[MSI]_year}^θ × {[ERI]_year}^(1-θ)where θ is equal to: {Fraction of influence of technological factors on the overall performance of the system; and, (1 − θ) = {Fraction of influence of economics on the overall performance of the system}.

Hence, the effective (RTPI)|_year, namely, [η_eff|_year] lies between the extrema limits of upper and lower Wiener-bounds; and, assuming an equally-weighted disposition of technology and economics, θ = 0.5 is suggested. Hence, a statistically-dictated probable value of RTPI in the presence of both technological and economics-related considerations corresponds to [η_eff|_year] with

θ = 0.5.

Appendix C

1.1 MSI and RTPI expressed via a set of relative performance indices (RPI)]

A brief listing of computed results on MSI and RTPI of incumbent mobile carriers indicated in [23] is depicted in Table 7

Table 7 Summary of computed results expressed in terms of a set of relative performance indices (RPI) denoting: (a) MSI; (b) RTPI; (c) Wiener upper-bound (UB) of RTPI and (d) Wiener lower-bound (LB) of RTPI pertinent to incumbent mobile carriers at Boston, MA: USA during the period 2010 through 2013