Page 24 - ITUJournal Future and evolving technologies Volume 2 (2021), Issue 1

ITU Journal on Future and Evolving Technologies, Volume 2 (2021), Issue 1

0.025 0.1

0.02 0.095

0.015 0.09

0.01 0.085

0.005 0.08

0 0.075
1 2 3 4 5 6 1 2 3 4 5 6

(a) Packet deadline: m = 10. (a) Packet deadline: m = 10.
10 -3 0.1
7
0.099
6
0.098
5
0.097
4
0.096
3 0.095
2 0.094

0.093
1
0.092
0 1 2 3 4 5 6
1 2 3 4 5 6
(b) Packet deadline: m = 30.
(b) Packet deadline: m = 30.
Fig. 4 – DPC vs LDF. Throughput comparison for different values of
Fig. 3 – DPC vs LDF. Drop rate comparison for different values of packet packet deadline m.
deadline m.
LDF algorithm in terms of drop rate as the number of 0.25
minimum‑throughput requirements increases. In Fig. 3b,
the deadline for the packets of user 1 is m = 30. We ob‑ 0.2
serve that the performance of LDF has been improved.
However, the DPC outperforms LDF in this case as well. 0.15
We observe that LDF is more sensitive on the size of the
deadline of the packets. 0.1
In Fig. 4, we compare the performance of the algo‑
rithm regarding the average total throughput of users 0.05
with minimum‑throughput requirements. In Fig. 4a, we
show results for packets deadline that is m = 10. Also in 0
this case, we observe that the DPC algorithm outperforms 0 0.5 1 1.5 4 2
the LDF. However, for larger deadlines, the performance 10
of the LDF is improved, as shown in Fig. 4b. Fig. 5 – DPC vs LDF. Min‑throughput requirements convergence. Pack‑
In Fig. 5, we provide results for the convergence time ets deadline: m = 100 slots.
of throughput requirements of one user. In this set‑up, user 1. We observe that the DPC algorithm converges
we consider a system with one user with packets with much earlier than the LDF algorithm. However, both algo‑
deadlines and six users with minimum‑throughput re‑ rithms converge after many slots. This explains the phe‑
quirements. In the previous results that are shown in nomenon of worst performance of LDF for small values of
Figures. 4 and 3, we observe that as we increase the m. Since the LDF algorithm allocates power to the users
value of m, we get a better performance of the LDF al‑ with largest throughput‑debt, it does not consider the re‑
gorithm. In the system of which the results are shown maining slots for each packet. Therefore, when the dead‑
in Fig. 5, we set a very large value to the deadlines of line, m, is small, the packets expire before the algorithm

8 © International Telecommunication Union, 2021