Quality of Service (QoS)

QoS is the ability of a network to give required bandwidth, delay, jitter and loss guarantees to different kinds of traffic. Main techniques:

1. Scheduling

Decides the order in which queued packets are sent.

• FIFO: first come, first served; simple, no differentiation
• Priority queuing: higher-priority queues are served first (risk of starving low priority)
• Weighted Fair Queuing (WFQ): each class gets a share of bandwidth by weight

2. Traffic shaping

Controls the rate at which traffic enters the network.

• Leaky bucket: constant output rate
• Token bucket: allows bursts up to the saved tokens

3. Admission control

A router or switch accepts a new flow only if it can meet the flow's QoS request, otherwise it rejects it.

4. Resource reservation

Bandwidth, buffer and CPU are reserved in advance for a flow along its path, e.g. with RSVP (Integrated Services model).

5. Other techniques

• Over-provisioning: more capacity than needed
• Buffering at the receiver to reduce jitter (streaming)
• Packet classification and marking with DiffServ (DSCP) so routers treat classes differently

Technique	Purpose
Scheduling	Fair/priority use of bandwidth
Traffic shaping	Smooth bursts, prevent congestion
Admission control	Avoid overload
Resource reservation	Guaranteed service