Beyond QCI: Understanding Performance Competition Between an MVNO and Its Host MNO

Abstract

Mobile virtual network operators (MVNOs) provide service over host mobile network operators (MNOs), yet users often assume that host carriers receive better performance. This paper presents, to our knowledge, the first simultaneous measurement study of MVNO-host MNO throughput competition with radio-side visibility. Across three operator families, five competition scenarios, and three U.S. cities, we run simultaneous TCP throughput tests over commercial subscriptions and collect cross-layer indicators including QCI, PCI, RAT mode, SA/NSA mode, resource block allocation, MCS, and MAC layer NR/LTE throughput. Our results show that host advantage is conditional rather than universal, and that no single factor consistently explains host-MVNO throughput ordering. Across operator families, observed performance differences are associated with different combinations of QoS class, RAT attachment, serving-cell context, radio-resource allocation, frequency band, and SA/NSA architecture. Under identical radio contexts, QoS-class ordering often coincides with downlink performance ordering. In contrast, when the radio contexts are different, RAT mismatch, cell attachment, or NSA LTE aggregation dominates the observed outcome. City-level comparisons further show that operator ordering can change across locations as RAT attachment, serving-cell selection, bandwidth, and radio conditions vary. Thus, host-MVNO performance must be interpreted through the joint radio and access context rather than from operator identity or any single QoS indicator.