Page 39 - ITUJournal Future and evolving technologies Volume 2 (2021), Issue 1

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

0.11 0.07
Reconfiguration - ST τ = 2
0.1 0.06 Reconfiguration - ST τ = 3
Reconfiguration - ST τ = 4
Signaling Overhead (Mbps)
Reconfiguration - ST τ = 5
0.09 0.05 No Reconfiguration - ST τ = 2
Max Packet Delay (ms) 0.07 Reconfiguration - ST-Max τ = 2 0.04 No Reconfiguration - ST τ = 5
No Reconfiguration - ST τ = 3
0.08
No Reconfiguration - ST τ = 4
0.06
0.03
0.05
Reconfiguration - ST-Max τ = 3
0.04
Reconfiguration - ST-Max τ = 4
Reconfiguration - ST-Max τ = 5
0.03
0.01
No Reconfiguration - ST-Max τ = 3
0.02 No Reconfiguration - ST-Max τ = 2 0.02
No Reconfiguration - ST-Max τ = 4
No Reconfiguration - ST-Max τ = 5
0.01 0
2 4 6 8 10 12 14 16 18 20 2 4 6 8 10 12 14 16 18 20
Stream Mean Rate π (Streams/Second) Stream Mean Rate π (Streams/Second)
Fig. 7 – Centralized Unidirectional Topology: Maximum packet delay for Fig. 9 – Centralized Unidirectional Topology: Stream average signaling
TAS with centralized con iguration (CNC) management. Overhead for TAS with centralized con iguration (CNC) management.
100
Reconfiguration ST τ = 2 2.5 Reconfiguration - ST τ = 2
90 Reconfiguration ST τ = 3 Reconfiguration - ST τ = 3
Reconfiguration ST τ = 4
Reconfiguration ST τ = 5 Reconfiguration - ST τ = 4
Reconfiguration - ST τ = 5
80 No Reconfiguration - ST τ = 2 2 No Reconfiguration - ST τ = 2
No Reconfiguration - ST τ = 3 No Reconfiguration - ST τ = 3
70
No Reconfiguration - ST τ = 4
No Reconfiguration - ST τ = 4
% Admission 60 Throughput (Gbps) 1.5
No Reconfiguration - ST τ = 5
No Reconfiguration - ST τ = 5
50
1
40
30 0.5
20
10 0
2 4 6 8 10 12 14 16 18 20 2 4 6 8 10 12 14 16 18 20
Stream Mean Rate π (Streams/Second) Stream Mean Rate π (Streams/Second)
Fig. 8 – Centralized Unidirectional Topology: Stream Admission per‑ Fig. 10 – Centralized Unidirectional Topology: ST total average through‑
centage for TAS with centralized con iguration (CNC) management. put measured at the sink for TAS with centralized con iguration (CNC)
management.
procedure provide a prescribed bandwidth share of the
egress port using time division multiplexing. With our from the data plane within the TSN domain, the delay is
empirically chosen parameters, the maximum delays is constant (around 4 s) throughout the simulation run.
capped to approximately 100 s which is suitable for the Stream registration and reservation introduce some con‑
considered topology and time‑critical ST traf ic that re‑ trol plane overhead. Fig. 9 shows the signaling overhead.
quires less than 1 ms of delay. More speci ically, the overhead is measured as the signal‑
While QoS metrics are important, another factor that de‑ ing traf ic rate in Mbit/s at the CNC for both incoming and
termines the performance gains is the admission ratio outgoing control (CDT) traf ic. Generally, the recon igu‑
for the system. Fig. 8 shows the stream admission ratio ration introduces more signaling overhead; however, Eth‑
for both recon iguration and no recon iguration. In gen‑ ernet generally has large bandwidths, thus the CDT traf ic
eral, each generated stream needs a data rate of about rates are minuscule compared to the link capacities. Fur‑
11.5 Mbps for a 50 s CT (which corresponds to approx‑ thermore, when = 2, we observe higher signaling over‑
imately 45 s of maximum ST slot size since we permit ST head due to accepting larger numbers of streams (rejec‑
traf ic to take up at most 90% of the CT) for each egress tions are inexpensive compared to acceptance) both with
port on the stream’s path with 1 packet injected by an and without recon iguration.
ST per CT and a ixed packet size of 64 B. With an egress Fig. 10 shows the average throughput measured at the
port channel capacity of = 1 Gbps, approximately sink for ST traf ic. We observe from Fig. 10 that the recon‑
86 streams can be accommodated. Compared to the “no iguration substantially increases the throughput com‑
recon iguration” approach, the recon iguration approach pared to the no recon iguration scenario. Typically, the
signi icantly improves the admission rates at the expense throughput is more than doubled by the recon iguration.
of higher BE traf ic delays, since the ST slot borrows BE To examine the reliability performance, Fig. 11 shows the
time slots to accommodate the ST streams. We also note BE packet loss ratio for mid and high BE traf ic loads ;

that increasing the maximum ST allocation above 90% we omitted the low BE traf ic load which has negligible
would increase the ST stream admission ratio, at the ex‑ losses. Since the CNC manages only ST streams, the TSN
pense of starving the BE traf ic. guarantees (which include zero packet loss since retrans‑
CDT traf ic that requests transmission guarantees from missions are in general too expansive for ST traf ic) are
the CNC experiences some delay before being either ad‑ only valid for ST streams. As the ST traf ic load increases
mitted or rejected. Since the control plane is out‑of band in the recon iguration scenario, the BE packet loss in‑

© International Telecommunication Union, 2021 23