📡Systems Approach to Computer Networks Unit 4 – Delay and Loss in Packet Networks

Key Concepts

• Delay refers to the time it takes for a packet to travel from its source to its destination across a network
• Loss occurs when packets fail to reach their intended destination due to various factors such as congestion or link failures
• Latency is the total time it takes for a packet to be transmitted, propagated, processed, and received
• Bandwidth is the maximum rate at which data can be transmitted over a network link and affects the overall delay
• Throughput represents the actual rate at which data is successfully transmitted and received over a network
• Jitter is the variation in delay between packets arriving at their destination which can impact real-time applications
• Quality of Service (QoS) refers to the ability of a network to provide different levels of service to different types of traffic
  • Prioritizes certain types of traffic (voice, video) over others (file transfers) to minimize delay and loss for critical applications

Causes of Delay and Loss

• Limited bandwidth on network links leads to congestion when traffic exceeds the available capacity
• Queuing delay occurs when packets are temporarily stored in buffers at network devices while waiting to be transmitted
• Propagation delay is the time it takes for a signal to travel from the sender to the receiver over a physical medium (copper wire, fiber optic cable)
• Processing delay happens as network devices (routers, switches) examine packet headers and determine where to forward them
• Transmission delay is the time required to push all the bits of a packet onto the link at the available bandwidth
• Packet loss can result from congestion when buffers at network devices become full and start dropping packets
• Link failures due to physical damage, hardware malfunctions, or configuration errors can cause packets to be lost in transit
  • Redundant paths and fault-tolerant designs help mitigate the impact of link failures on packet loss

Types of Network Delay

• End-to-end delay is the total time it takes for a packet to travel from its source to its destination across the network
• Queuing delay occurs when packets are temporarily stored in buffers at network devices while waiting to be transmitted
  • Influenced by factors such as link bandwidth, traffic load, and queuing disciplines (FIFO, priority queuing)
• Propagation delay is the time it takes for a signal to travel from the sender to the receiver over a physical medium
  • Depends on the distance between the sender and receiver and the propagation speed of the medium (copper wire, fiber optic cable)
• Processing delay happens as network devices examine packet headers and determine where to forward them
  • Affected by the processing power and efficiency of the network devices (routers, switches)
• Transmission delay is the time required to push all the bits of a packet onto the link at the available bandwidth
  • Calculated as the packet size divided by the link bandwidth ($delay = \frac{packet\_size}{bandwidth}$)
• Serialization delay is the time it takes to convert a packet from parallel to serial format for transmission over a link
  • Becomes significant for high-bandwidth links and large packet sizes

Measuring Network Delay

• Round-trip time (RTT) is the time it takes for a packet to travel from a sender to a receiver and back again
  • Measured using tools like

    ping

    which sends an ICMP echo request and waits for an echo reply
• One-way delay (OWD) is the time it takes for a packet to travel from a sender to a receiver in one direction
  • More challenging to measure accurately due to clock synchronization issues between the sender and receiver
• Delay variation or jitter is the difference in delay between successive packets in a flow
  • Important for real-time applications (voice, video) that require consistent packet arrival times
• Network Time Protocol (NTP) helps synchronize clocks across devices in a network to enable accurate delay measurements
• Passive measurement techniques involve observing existing traffic without injecting additional packets into the network
  • Can use tools like

    tcpdump

    or

    wireshark

    to capture and analyze packet timestamps
• Active measurement techniques involve injecting test packets into the network and measuring their delay and loss
  • Examples include

    ping

    ,

    traceroute

    , and

    iperf

    which generate specific types of traffic for measurement purposes

Packet Loss: Causes and Effects

• Congestion is a primary cause of packet loss when network links become overloaded with traffic
  • Occurs when the arrival rate of packets exceeds the available bandwidth or processing capacity of network devices
• Buffer overflow happens when the queues at network devices fill up and start dropping incoming packets
  • Can be mitigated through proper buffer sizing and active queue management techniques (Random Early Detection)
• Link failures due to physical damage, hardware malfunctions, or configuration errors can cause packets to be lost in transit
  • Redundant paths and fault-tolerant designs help maintain connectivity and minimize packet loss
• Packet corruption can occur due to noise, interference, or hardware issues leading to unrecoverable errors and discarded packets
  • Error detection and correction mechanisms (checksums, forward error correction) help identify and recover from corrupted packets
• Packet reordering can happen when packets take different paths through the network and arrive out of sequence
  • Can trigger duplicate acknowledgments and retransmissions in TCP leading to reduced throughput and increased delay
• Packet loss has a significant impact on the performance of network applications and protocols
  • Triggers retransmissions and congestion control mechanisms in TCP resulting in reduced throughput and increased delay
  • Degrades the quality of real-time applications (voice, video) that are sensitive to missing or delayed packets

Congestion Control Mechanisms

• Congestion control aims to prevent and mitigate network congestion by regulating the rate at which senders inject packets into the network
• TCP congestion control algorithms (Reno, Cubic) dynamically adjust the sender's transmission rate based on network feedback
  • Slow start gradually increases the transmission rate until congestion is detected through packet loss or increased delay
  • Congestion avoidance maintains a stable transmission rate near the network's capacity while probing for additional bandwidth
• Active queue management techniques (Random Early Detection) proactively drop packets before buffers become full
  • Provides early feedback to senders about impending congestion allowing them to reduce their transmission rates
• Explicit congestion notification (ECN) allows network devices to mark packets to indicate congestion instead of dropping them
  • Enables senders to react to congestion more quickly and avoid unnecessary packet loss and retransmissions
• Fair queuing algorithms (Weighted Fair Queuing) allocate bandwidth fairly among competing flows to prevent starvation
  • Ensures that no single flow can dominate the link and cause excessive delay or loss for other flows
• Traffic shaping and policing mechanisms enforce bandwidth limits and drop excess packets to prevent congestion
  • Helps maintain a consistent quality of service for critical applications and prevent network overload

Quality of Service (QoS) Strategies

• QoS aims to provide differentiated treatment for different types of network traffic based on their requirements and priorities
• Classification and marking techniques identify and label packets according to their QoS requirements (delay, loss, bandwidth)
  • Can use fields in the packet header (IP ToS, DSCP) or application-layer information (port numbers) for classification
• Queuing and scheduling algorithms prioritize and manage the transmission of packets based on their QoS classes
  • Priority queuing gives strict preference to high-priority traffic (voice, video) over low-priority traffic (file transfers)
  • Weighted fair queuing allocates bandwidth proportionally among different QoS classes ensuring fairness and preventing starvation
• Traffic policing and shaping enforce bandwidth limits and drop or delay excess packets to maintain QoS guarantees
  • Policing drops packets that exceed the specified rate limit while shaping buffers them and smooths out bursts
• Admission control mechanisms limit the amount of traffic admitted into the network based on available resources and QoS requirements
  • Prevents network overload and ensures that QoS guarantees can be met for admitted flows
• Resource reservation protocols (RSVP) allow applications to request and reserve network resources for their QoS needs
  • Enables end-to-end QoS provisioning across multiple network domains and devices
• QoS-aware routing protocols (QoS-OSPF) consider QoS metrics (delay, loss, bandwidth) in addition to traditional routing metrics (hop count)
  • Helps find paths that meet the QoS requirements of different types of traffic and avoid congested or low-performance links

Real-world Applications and Case Studies

• Voice over IP (VoIP) and video conferencing applications require low delay, jitter, and loss to maintain acceptable quality
  • Employ QoS mechanisms (priority queuing, traffic shaping) to prioritize voice and video packets over other traffic types
• Online gaming and virtual reality applications have strict requirements for low latency and consistent packet delivery
  • Use UDP for fast packet transmission and implement client-side prediction and interpolation to compensate for network delays
• Streaming video services (Netflix, YouTube) adapt the video quality based on available bandwidth and network conditions
  • Employ adaptive bitrate streaming techniques (DASH, HLS) to dynamically adjust the video resolution and minimize buffering and stalls
• Software-defined networking (SDN) and network function virtualization (NFV) enable flexible and dynamic QoS provisioning
  • Allow network operators to program and orchestrate QoS policies across multiple network devices and services
• Content delivery networks (CDNs) reduce delay and improve performance by caching content closer to the users
  • Employ geographically distributed servers and intelligent routing mechanisms to serve content from the optimal location
• Satellite and wireless networks face unique challenges in terms of delay, loss, and bandwidth variability
  • Implement specialized protocols (TCP-Spoofing, Performance Enhancing Proxies) to optimize performance over high-latency and error-prone links
• Internet of Things (IoT) applications have diverse QoS requirements ranging from low-power sensors to mission-critical control systems
  • Employ lightweight protocols (MQTT, CoAP) and edge computing architectures to minimize delay and loss for IoT devices