Proportionally fair

For the resource allocation problem, see Proportional division.

Proportional fair is a compromise-based scheduling algorithm. It is based upon maintaining a balance between two competing interests: Trying to maximize total [wired/wireless network] throughput while at the same time allowing all users at least a minimal level of service. This is done by assigning each data flow a data rate or a scheduling priority (depending on the implementation) that is inversely proportional to its anticipated resource consumption.^[1] ^[2]

Weighted fair queuing

Proportionally fair scheduling can be achieved by means of weighted fair queuing (WFQ), by setting the scheduling weights for data flow $i$ to $w_{i}=1/c_{i}$ , where the cost $c_{i}$ is the amount of consumed resources per data bit. For instance:

In CDMA spread spectrum cellular networks, the cost may be the required energy per bit in the transmit power control (the increased interference level).
In wireless communication with link adaptation, the cost may be the required time to transmit a certain number of bits using the modulation and error coding scheme that this required. An example of this is EVDO networks, where reported SNR is used as the primary costing factor.
In wireless networks with fast Dynamic Channel Allocation, the cost may be the number of nearby base station sites that can not use the same frequency channel simultaneously, in view to avoid co-channel interference.

User prioritization

Another way to schedule data transfer that leads to similar results is through the use of prioritization coefficients.^[3] Here we schedule the channel for the station that has the maximum of the priority function:

$P={\frac {T^{\alpha }}{R^{\beta }}}$

$T$ denotes the data rate potentially achievable for the station in the present time slot.
$R$ is the historical average data rate of this station.
$\alpha$ and $\beta$ tune the "fairness" of the scheduler.

By adjusting $\alpha$ and $\beta$ in the formula above, we are able to adjust the balance between serving the best mobiles (the ones in the best channel conditions) more often and serving the costly mobiles often enough that they have an acceptable level of performance.

In the extreme case ( $\alpha =0$ and $\beta =1$ ) the scheduler acts in a round-robin fashion and serves all mobiles equally often, with no regard for resource consumption. If $\alpha =1$ and $\beta =0$ then the scheduler will always serve the mobile with the best channel conditions. This will maximize the throughput of the channel while stations with low $T$ are not served at all.^[2]

Using  $\alpha \approx 1$  and  $\beta \approx 1$  will yield the proportional fair scheduling algorithm used in 3G networks.^[3]

This technique can be further parametrized by using a "memory constant" that determines the period of time over which the station data rate used in calculating the priority function is averaged. A larger constant generally improves throughput at the expense of reduced short-term fairness.

References

↑ Kushner, H. J.; Whiting, P.A. (July 2004), "Convergence of proportional-fair sharing algorithms under general conditions", IEEE Transactions on Wireless Communications, 3 (4): 1250–1259, doi:10.1109/TWC.2004.830826. Cite uses deprecated parameter |coauthors= (help)
1 2 Guowang Miao, Jens Zander, Ki Won Sung, and Ben Slimane, Fundamentals of Mobile Data Networks, Cambridge University Press, ISBN 1107143217, 2016.
1 2 Yang, Ji; Zhang Yifan; Wang Ying; Zhang Ping (2004-11-29), "Average rate updating mechanism in proportional fair scheduler for HDR", IEEE Global Telecommunications Conference, 2004., 6, IEEE, pp. 3464–3466, doi:10.1109/GLOCOM.2004.1379010 Cite uses deprecated parameter |coauthors= (help)

Additional reading

Andrews, Matthew (September 2004), "Instability of the Proportional Fair Scheduling Algorithm for HDR", IEEE Transactions on Wireless Communications, 3 (5): 1422–1426, doi:10.1109/TWC.2004.833419.
Andrews, Matthew; K. Kumaran, K. Ramanan, A. Stoyar and Phil Whitting (February 2001), "Providing Quality of Service over a Shared Wireless Link", IEEE Communications, 39 (2): 150–154, doi:10.1109/35.900644. Cite uses deprecated parameter |coauthors= (help)
Parruca, Donald; M. Grysla , S. Goertzen, and J. Gross (September 2013), "Analytical Model of Proportional Fair Scheduling in Interference-Limited OFDMA/LTE Networks", Proc. of IEEE Vehicular Technology Conference: 1–7., doi:10.1109/VTCFall.2013.6692106 Cite uses deprecated parameter |coauthors= (help)

This article is issued from Wikipedia - version of the 11/2/2016. The text is available under the Creative Commons Attribution/Share Alike but additional terms may apply for the media files.

Proportionally fair

Weighted fair queuing

User prioritization

See also

References

Additional reading