There is a trade-off in network slicing between the twin goals of providing tailored performance and increasing resource utilisation through increased opportunities for sharing. To balance this trade-off, we propose a system consisting of assured resources, which are available if needed over the lifetime of a slice, and auxiliary resources, which are offered on a probabilistic basis in the short-term based on forecasted resource demand. We employ the practice of overbooking, which is widely used in many industries such as airlines and hotels, to increase resource utilisation when offering auxiliary resources. After deriving probabilistic results relating to the availability of auxiliary resources, we then design an algorithm to determine how much to overbook by. We also propose several ways of distributing auxiliary resource offers among slices, as well as providing an approach for dealing with overbookings. Finally, we highlight the conditions that are conducive to effective overbooking by examining the effects of varying several key system parameters, providing both analytical and numerical results.

In this paper, we explore the idea that 5G will permit the use of multiple waveforms, with each service employing a waveform that is best suited for it. We look at a 5G machine-type communication (MTC) scenario consisting of clustered user equipment employing device-to-device (D2D) communication, such as a smart factory with intercommunicating machinery. The overhead associated with synchronizing a large number of machine-type D2D user equipment (DUE) comes at a cost that may render synchronous communication infeasible or undesirable. Based on this motivation, we consider multiple possible combinations of prominent 5G waveform candidates for cellular users and DUEs, examining the asynchronous performance of all waveforms under consideration and using the performance of synchronous orthogonal frequency division multiplexing (OFDM) as a baseline for comparison. Specifically, we focus on the coexistence of waveforms in which the ordinary cellular users employ OFDM for synchronous communication, as in LTE, and the machine-type DUEs, operating asynchronously, employ a different waveform. When DUEs employ filter bank multicarrier with offset-QAM, the average achieved rate is marginally greater than the synchronous OFDM baseline case, and approximately 43% greater than the asynchronous OFDM case. This result is encouraging, as the benefits of asynchronous D2D communication could be enjoyed in MTC scenarios without suffering any performance reduction compared to the synchronous OFDM scenario. We then investigate how the relative performance of different waveform choices depends on the scenario by varying key parameters. Notably, for asynchronous communication, increasing the transmit power of DUEs results in diminishing benefits unless the DUEs employ a waveform that mitigates interdevice leakage interference.