Application of Revenue Management to Satellite Communications

Abstract

As the demand for satellite Internet continues to grow, satellite communication (SatCom) operators are faced with the challenge of effectively managing their capacity sales. While Revenue Management (RM) techniques have been widely used in other industries such as airline, hotel, and car rental services, the application of these methods in the context of SatCom is still scarce. This Thesis aims to bridge this gap by developing RM concepts, techniques, and optimization algorithms specifically tailored to the unique operational characteristics of SatCom capacity management and sales. The proposed SatCom RM method guides operators with quantitative recommendations of the amount of capacity to sell to different products in time and in different regions to maximize revenues.

Though SatCom has characteristics that favor the use of RM concepts (perishable inventory, fixed capacity with a low variable cost, the possibility to segment demand), there are unique structural characteristics that complicate the development of SatCom RM models. The primary challenge is that different products consume varying amounts of capacity, with larger terminal size products utilizing less power on a satellite than smaller terminal size products. Moreover, the selling practices in SatCom are complex because products may be sold in one period and consumed across multiple periods in which additional sales are made. This requires rolling both the selling and consumption periods. Lastly, the SatCom RM problem poses a multidimensional network problem, as products can consume bundles of resources in both space and time.

We extend two commonly used airline RM algorithms, Expected Marginal Seat Revenue (EMSRb) heuristic and Displacement Adjusted Virtual Nesting (DAVN) to the SatCom problem to create booking limits. The booking limits recommend a threshold amount of capacity an operator should sell of each product. The contribution of this Thesis is the modification of established airline RM algorithms to handle products with variable capacity uptakes. Further, these algorithms typically account for displacement costs of products, but only in one dimension of space or time (e.g., selling an airline flight that uses multiple spatial legs may displace capacity away from flights that only use one leg). Our modifications allow for the consideration of displacement costs in both dimensions of space and time.

In order to evaluate the effectiveness of our inventory control approach, we conduct simulations of various demand scenarios and compare the revenue gains to a baseline scenario with no controls, as well as a simpler method that does not consider product duration. In a large-scale simulation spanning three years and encompassing thousands of product requests, we observe revenue gains ranging from 15%-30% depending on the demand scenario. Then, we extend the model to multiple zones and achieve 2%-10% revenue improvement using our Multi-Zone DAVN method compared to the DAVN method applied to each zone separately.