Full duplex communication has been well explored in the past decade, with several physical layer designs proposed for effectively doubling the throughput of mobile devices. However, these systems are far from being well-deployed in the Wi-Fi and cellular applications, where most of the research has been done. In this paper, we investigate some of the fundamental pragmatic challenges in deploying full duplex that explain the reluctance from industry players in deploying this feature in commercial end-user devices and base stations at a large scale. Upon doing so, we identify that the problems lie not quite at the radio layer -- but at layers below and above it. At the hardware layer, we find that the power and complexity of IC-implementations of full duplex far exceeds that of other existing technologies that achieve similar throughput gain, such as multi-user MIMO. We further identified how higher-layer cellular and Wi-Fi traffic patterns and MAC protocols are fundamentally ill-suited to the full duplex paradigm. We report empirical analysis from the power consumption of an IC-implementation of a full duplex cancellation circuit, demonstrating that full duplex cancellation circuitry would consume at least 57 percent more power than a MIMO system achieving identical throughput gains. We also show how current traffic patterns reduce the benefits of full duplex to 1.25X instead of the expected 2X.
Dinesh BharadiaEmily McMilinSachin Katti
Jin ZhouNegar ReiskarimianJelena DiakonikolasTolga DinçTingjun ChenGil ZussmanHarish Krishnaswamy