I NSPECCIÓN Y V IGILANCIA

QoS, in general, is related to the underlying data transport network and measures network-related parameters. Several metrics have been developed for the evaluation of QoS, including packet delay, loss, jitter, etc.

2.3.2.1 End-to-end Delay

End-to-end delay, as defined in [61], refers to the time taken for a packet to be transmit- ted across a network from source to destination. Delay in wireless networks compromises two primary parts: end-point delay and network delay.

End-point delay refers to the time taken before a piece of data is pushed to the net- work at the end points. It is mainly caused by the processing of information, such as data encoding and decoding which are used in multimedia streaming applications, or sample analysis process employed in Voice over IP (VoIP) applications.

Network delay specifies how long it takes for a piece of data to travel across the net- work from the time the first bit is pushed into the network to the time the last bit is received

by the destination. It is divided into several parts:

• Processing delay is the time taken to process a packet at intermediate nodes within a network. For example, routers may check for bit-level errors in the packet or look up the routing table to determine where the packet’s next destination is.

• Queuing delay is the time a packet spends in the queues, waiting to be transmitted. It is most often used in reference to routers, which can only process one packet at a time. If packets arrive faster than the router’s processing speed, they are put into the queue until being executed by the router.

• Transmission delay refers to the time it takes to push the all bits of a packet onto

the link. DT is calculated as the total number of bits in a packet divided by the

transmission rate, as in equation (2.2), where N is the number of bits, and R is the rate of transmission.

DT = N/R (2.2)

• Propagation delay is the time required to deliver a packet over a medium. It is com- puted as the ratio between the link length L and the propagation speed, S, over the specific medium, as shown in equation (2.3). Propagation delay varies for different medium. For instance, the value reaches 10 Gbps in fibre while in copper up to 15 Mbps of speed is obtained.

DP = L/S (2.3)

2.3.2.2 Packet Loss Rate

Packet loss rate, as specified in [62], is defined as the number of dropped packets di- vided by total number of packets transmitted by the source. The main reasons leading to packet loss are:

• Fading effect which is a characteristic of wireless data transmission. Signal strength experiences degradation over the network medium which could result in packet loss. • Interference among devices sharing the same frequency as an access point could lead

to packet loss and retransmission, which seriously affects user experience.

• Unmatched transmission speed, which means the outer link has a slower transmis- sion speed than the inner link in a node or access point. In this case, when packets arrive at the node, they cannot be forwarded without being delayed at the queue. When the available buffer space is fully filled, upcoming packets will be dropped. • Traffic fluctuation which may cause aggregation. Even if the outgoing bitrate is

similar or the same as the incoming bitrate of a node, packets might get dropped at the queue when traffic from different sources all arrive at one time and there is no enough room for all packets. Besides, data processing takes time and might become the bottleneck in this scenario.

2.3.2.3 Jitter

Jitter is the undesired deviation from true periodicity of an assumed periodic signal in electronics and telecommunications. In computer networks, jitter, also known as Packet Delay Variation (PDV), often refers to the variability over time of the packet latency across a network. It is mostly often measured as the difference of delay between successive pack- ets. The term is defined in [63] and can be caused by some primary reasons:

• Variation in packet scheduling time. For example, a multimedia transmission pro- cess has to contend for CPU time with other processes and hence there may be some transmit time jitter introduced by scheduling.

• Network congestion. Congestion in networks can hardly be predicted and this leads to varied delay in packet delivery.

work load, the traffic is sometimes routed over multiple routes. Jitter is introduced if the delays across different routes vary significantly.

• Internal load sharing. This approach is employed by some routers to provide a multi-processing approach in which packets are processed by multiple parallel queues. The short term differences in queue size could result in low levels of jit- ter.

Despite the major aspects, other factors such as routing table updates and time drifting could also lead to serious jitter.

2.3.2.4 QoS for Multimedia Services

Although the metrics such as packet delay, jitter, and loss are commonly used for vari- ous types of applications, the requirements asked by each application service varies signif- icantly. For example, applications such as email and file transfer allow longer delays and do not care about jitter levels much, while multimedia streaming services are much more sensitive to packet delay or jitter.

Table 2.4 lists the network performance objectives for IP-based applications, as sug- gested in [64]. QoS levels are categorized into 6 classes, with class 0 representing the best quality while class 5 indicating the worst. Class 0 and 1 can be used for real time data tranmission such as video coferencing and VoIP, while the other classes are more suitable for transaction data transmission, such as file transfer over the networks.