**3.2 QoS applied to multimedia TCP flows (no min/max bandwidth set)**

The second campaign aimed at verifying the efficiency of the QoS management server. The Iperf was moved in the (C) node, so as to check Link 3.

An ethernet cable subsituted the wireless connection during a PPP connection. Delays and packet loss, in fact, are not particularly relevant in this kind of control, attention being mainly focused on flow management.

The following tools were adopted: (1) QoS Server; (2) Server-side Vlc for MMS over http video flow transmission (www.videolan.org/vlc); (3) Client-side Vlc for flow reception; (4) WireShark.

Fig. 7, diagrammed through WireShark, represents the scenario and the first measurements of this campaign. It refers to the following profile: no QoS applied, symmetric upload and dowmlod of 1Mbit/s, no bandwidth guaranteed. The client receives the first video (red line) until second 230, then the second video (green line) starts and the firts one is interrupted at second 280.

As expected, in the concurrent period (sec. 230 to 280) both TCP videos are blocked, the bandwidth being inadequate to support both of them.

Fig. 8 represents the second test and involves three videos on three distinct ports; in this case, a QoS profile was enabled which guarantees an increasing priority from the first to the third flow.

First starts the video on the lowest priority port (blue curve); the intermediate priority video starts 20 seconds later (green curve) and, in consequence, the first data flow declines. In the period between sec. 40 to 80 the maximum throughput was increased, so as to emphasize the effect of dark and still scenes in the second video. In this way, the total throughput is constant and bandwidth waste is kept under control.

A Testbed About Priority-Based Dynamic

priority flow starts (red curve).

such method and Fig. 10 reports the results.

Fig. 9. TCP concurrent flows with QoS (bit/s).

totally flattened, but only slowed down.

can not reach the maximum speed.

**Class Priority level Min. bandwidth %** 

1 Highest priority 0% 100% 2 Intermediate 30% 30% 3 Lowest priority 20% 100%

ones.

flows.

Connection Profiles in QoS Wireless Multimedia Networks 87

Nevertheless, this kind of priority management among traffic classes implies the almost complete cancellation of the less important services for the benefit of the most important

In the initial QoS profile, an increasing priority was assigned to the three TCP flows on distinct ports. Fig. 9 shows the measures obtained. The less priority flow (blue curve) is strongly limited by the second one (green curve). They are both flattened when the highest

In order to improve such results, each traffic class was then assigned a minimum and a maximum throughput (Tab. 3). The third campaign tries to demonstrate the effectiveness of

**guaranteed** 

Table 3. minimum throughput guaranteed and maximum available, as assigned to single

Note that an upper bound having been imposed to the intermediate flow, the lowest is not

The most priority flow starts at second 50 and, in consequence, the less priority traffic becomes slower, but not more than the 20% guaranteed. The same applies to the intermediate flow, for which a 30% at least is available. The highest priority flow, of course,

**Max. bandwidth % at disposal** 

Fig. 7. two concurrent video flows without QoS (bit/s): scenario and results.

The same observations apply to the third and highest priority flow (red curve): the three videos share the bandwidth and, thanks to QoS, the lowest priority one is flattened, the medium is assigned less bandwidth and the highest has the best quality.

Fig. 8. three concurrent videos with QoS (bit/s).

#### **3.3 Bandwidth control in QoS management**

As the previous measurements showed, the QoS policy adopted helps to avoid bandwidth waste and guarantees a better service, especially for VoIP, IPTV support etc.

Fig. 7. two concurrent video flows without QoS (bit/s): scenario and results.

medium is assigned less bandwidth and the highest has the best quality.

Fig. 8. three concurrent videos with QoS (bit/s).

**3.3 Bandwidth control in QoS management** 

The same observations apply to the third and highest priority flow (red curve): the three videos share the bandwidth and, thanks to QoS, the lowest priority one is flattened, the

As the previous measurements showed, the QoS policy adopted helps to avoid bandwidth

waste and guarantees a better service, especially for VoIP, IPTV support etc.

Nevertheless, this kind of priority management among traffic classes implies the almost complete cancellation of the less important services for the benefit of the most important ones.

In the initial QoS profile, an increasing priority was assigned to the three TCP flows on distinct ports. Fig. 9 shows the measures obtained. The less priority flow (blue curve) is strongly limited by the second one (green curve). They are both flattened when the highest priority flow starts (red curve).

In order to improve such results, each traffic class was then assigned a minimum and a maximum throughput (Tab. 3). The third campaign tries to demonstrate the effectiveness of such method and Fig. 10 reports the results.

Fig. 9. TCP concurrent flows with QoS (bit/s).


Table 3. minimum throughput guaranteed and maximum available, as assigned to single flows.

Note that an upper bound having been imposed to the intermediate flow, the lowest is not totally flattened, but only slowed down.

The most priority flow starts at second 50 and, in consequence, the less priority traffic becomes slower, but not more than the 20% guaranteed. The same applies to the intermediate flow, for which a 30% at least is available. The highest priority flow, of course, can not reach the maximum speed.

A Testbed About Priority-Based Dynamic

redistributed.

years.

of the most relevant ones.

Connection Profiles in QoS Wireless Multimedia Networks 89

The two clients share the same profile, so the bandwidth is equally divided and

At second 150 the third client (red curve) connects to the system; the minimum throughput of his or her profile is four times the others', so clients 1 and 2 are limited accordingly.

When client 1 disconnects, bandwidth is distributed among clients 2 and 3 in a ratio of 1 to 4.

Finally, client 3 logs out and the whole bandwidth is at client's 2 disposal.

Fig. 11. bandwidth proportional reassignment among connected users.

Smart Navigation and the braoader field of Infomobility.

**4. The PEGASUS project: Real time support in infomobility services** 

2050, nearly 70% of the global population will be living in larger cities [23].

One of the most important applications to which QoS techniques have been applied are

Transportation is one of the main fields where advanced technological systems can improve human life in a significant way: risks due to accidents, time wasted travelling and pollution could be highly reduced by applications for vehicle localization, behaviour prediction, etc. These considerations are at the basis of the increasing interest that ITS are gaining in these

Furthermore, the latest study on global urbanization conducted by the Population Division of the Department of Economic and Social Affairs of the United Nations predicts that, in

This immense aggregation of people will surely pose great challenges to the sustainability of modern lifestyle, and the problem of an efficient management of mobility stands out as one

As this measure shows, if guaranteed bandwidth percentages are properly managed, a high QoS can be obtained in an easy and immediate way.

Fig. 10. TCP concurrent flows with bounded bandwidth QoS (bit/s).

### **3.4 Proportional reassignment of bandwidth**

An important feature that must be handled in this kind of QoS management is bandwidth reassignment proportionally to each user's minimum guaranteed.

In this case, three PCs and the usual tools were adopted and two kinds of profiles were defined (Tab. 4).


Table 4. Minimum and maximum throughput for each client.

Clients were connected to the server through the PPPoE protocol; the maximum throughput between clients and server was set to 3Mbit/s, so as to simulate a set of wireless relays.

In this case, the upper bandwidth was to be shared among concurrent users and, in consequence, none was to reach the maximum.

Results are shown in Fig. 11: initially, the only connected client 1 (green curve) gets the whole bandwidth available according to his profile (3 Mbit/s).

After 100 seconds approximately, client 2 is also connected, throughput is assigned according to the minimum guaranteed and exceeding bandwitdh is reassigned according to such value.

As this measure shows, if guaranteed bandwidth percentages are properly managed, a high

QoS can be obtained in an easy and immediate way.

Fig. 10. TCP concurrent flows with bounded bandwidth QoS (bit/s).

reassignment proportionally to each user's minimum guaranteed.

Clients 1 and 2 128 Kbit/s 3 Mbit/s Client 3 600 Kbit/s 3 Mbit/s

Table 4. Minimum and maximum throughput for each client.

whole bandwidth available according to his profile (3 Mbit/s).

An important feature that must be handled in this kind of QoS management is bandwidth

In this case, three PCs and the usual tools were adopted and two kinds of profiles were

**Clients Min. guaranteed Max. at disposal** 

Clients were connected to the server through the PPPoE protocol; the maximum throughput between clients and server was set to 3Mbit/s, so as to simulate a set of wireless relays.

In this case, the upper bandwidth was to be shared among concurrent users and, in

Results are shown in Fig. 11: initially, the only connected client 1 (green curve) gets the

After 100 seconds approximately, client 2 is also connected, throughput is assigned according to the minimum guaranteed and exceeding bandwitdh is reassigned according to

**3.4 Proportional reassignment of bandwidth** 

consequence, none was to reach the maximum.

defined (Tab. 4).

such value.

The two clients share the same profile, so the bandwidth is equally divided and redistributed.

At second 150 the third client (red curve) connects to the system; the minimum throughput of his or her profile is four times the others', so clients 1 and 2 are limited accordingly.

When client 1 disconnects, bandwidth is distributed among clients 2 and 3 in a ratio of 1 to 4.

Finally, client 3 logs out and the whole bandwidth is at client's 2 disposal.

Fig. 11. bandwidth proportional reassignment among connected users.
