Chapter 5
HSDPA Performance Measurements
Most1 of the work published on High-Speed Downlink Packed Access (HSDPA) during recent years concentrates on system-level simulations [27, 28, 32, 33, 45, 48, 49, 51, 57] in which an analytical model [2, 3] of the physical layer is employed. Other theoretical studies focusing entirely on simulations look into specific details of the HSDPA physical layer such as, for example, Hybrid Automatic Repeat reQuest (HARQ) [6], receive antenna diversity [29], equalizer architectures [18, 34], radio-frequency hardware impairments [58], or link adaptation [47]. A meaningful assessment of the performance, however, requires the evaluation of the complete physical layer rather than the evaluation of individual physical layer parts. Such an evaluation was, for example, carried out in [36], which presents simulation results of a complete Single-Input Single-Output (SISO) HSDPA system including link adaptation. Furthermore, [36] shows a comparison between the simulated throughput and the Shannon capacity of the Additive White Gaussian Noise (AWGN) channel. Since HSDPA is typically operated in frequency-selective fading channels, however, the AWGN channel capacity, which only considers the mean Signal to Noise Ratio (SNR), is not a good performance bound. A much better, more realistic, and tighter bound can be obtained by calculating the channel capacity based on the channel realizations of the simulation.
Apart from simulations, HSDPA has been evaluated ...