On asymptotically fair transmission scheduling over fading channels with measurement delay

We examine the effect of measurement delays on the throughput of certain downlink data packet systems operating over Rayleigh fading channels. In these systems the base station (BS) schedules a single user at a time and transmits at a rate which is based on measurements of the BS's received power performed by the user. The same subject was also discussed in a recent paper, titled "Asymptotically fair transmission scheduling over fading channels" by Berggren and Jantti (2004). They studied the effects of measurement delays on the operation of the scheduler. As a consequence of the delay, it is possible that by the time the BS actually transmits data, the propagation channels might have changed, such that the scheduler's choice no longer conforms to the desired scheduling policy, resulting in loss of system throughput and multi-user gain. Berggren and Jantti assume that perfect link adaptation is nevertheless always assured, i.e., that the BS transmits at a rate that matches precisely the current state of the selected user's channel. However, measurement delays lead also to mismatch between the data rate the BS transmits to the chosen user, and the current state of the propagation channel to that user. We believe that this mismatch must also be taken into account. In this letter we propose to use a backoff factor and analyze its mitigating effect on losses introduced by measurements delay in a Rayleigh fading environment. We then show the resulting gain in system throughput for optimal choices of the backoff factor