DiffServ, Multicast, RTP, SCTP, SIP, H.323 & Watermarking

DiffServ Expedited and Assured Forwarding

Two PHBs (Per-Hop Behaviors) are defined for DiffServ:

  • Expedited Forwarding PHB: Applied to real-time traffic and related to the guaranteed service transfer capability. It specifies that the departure rate of a class of traffic from a router must equal or exceed a configured rate.
  • Assured Forwarding PHB: Applied to elastic traffic and related to the controlled load service transfer capability. It divides traffic into four classes, each guaranteed a minimum amount of bandwidth and buffering.

IP Multicast vs. CDN

There are major differences between IP multicast and Application-Level Multicast (ALM) as deployed in CDNs:

  • Multicast Group Identifier:
    • IP Multicast: A class-D address identifies a multicast group.
    • CDN: A URL or other application-related key identifies a multicast group.
  • Multicast Group Members:
    • IP Multicast: Users explicitly subscribe to certain multicast content via the IGMP protocol.
    • CDN: Requests are redirected to the user’s local CDN server.
  • Network Topology:
    • IP Multicast: Routers exactly reflect the physical network topology.
    • CDN: Servers form a logical overlay network on top of the physical network infrastructure.
  • Multicast Routing:
    • IP Multicast: Routing relies on the underlying unicast routing protocols and employs simple metrics (e.g., number-of-hops or delay).
    • CDN: Routing is much simpler (e.g., minimum-delay-path spanning tree, widest-path spanning tree, etc.).

Interdomain Multicast Routing Protocols

When group members are spread across different domains (AS), an interdomain multicast routing protocol is needed. Examples include:

  • Multicast Border Gateway Protocol (MBGP): An extension of BGP and a shared-group multicast routing protocol.
  • Multicast Source Discovery Protocol (MSDP)
  • Border Gateway Multicast Protocol (BGMP)

RTP and RTCP: Real-Time Interactive Protocols

Fulfillment of requirements with RTP and RTCP:

  • Sender-Receiver Negotiation: Cannot be satisfied directly; another protocol (SIP) must be considered.
  • Creation of Packet Stream: Each data chunk is encapsulated in an RTP packet with a sequence number.
  • Source Synchronization: Satisfied by an identifier and relative timestamp in the RTP packet, and the absolute timestamp in the RTCP packet.
  • Error Control: Using the sequence number in the RTP packet, the application can regenerate lost packets using FEC methods.
  • Congestion Control: Achieved by feedback from the receiver using RTCP receiver report packets, allowing the sender to adjust compression algorithms.
  • Jitter Removal: Achieved by the timestamp and sequence number in each RTP packet to buffer playback.
  • Sender Identification: Achieved by the CNAME included in RTCP packets.

SCTP Features and Services

Stream Control Transmission Protocol (SCTP) features and services include:

  • Message-Oriented Protocol: SCTP transports a sequence of messages, rather than a stream of bytes like TCP. Similar to UDP, SCTP sends a message in one operation, passed to the receiving application process in one operation.
  • Connection-Oriented Service: Like TCP, SCTP is connection-oriented, using an association negotiated before sending data.
  • Message Ordering: Optional; applications can process messages in the order of receipt instead of the order of sending, allowing urgent data processing.
  • Full-Duplex Communication: Like TCP, data can flow in both directions simultaneously.
  • Reliable Service: Like TCP, it’s a reliable transport protocol using an acknowledgment mechanism.
  • Path Selection and Monitoring: SCTP selects and monitors a primary data transmission path and tests connectivity.
  • Security: Improves security against DoS attacks, such as SYN attacks.

SIP Services

SIP provides the following services:

  • Call Establishment:
    • Finds the location of users (their IP addresses).
    • Determines if users are able or willing to participate in the conference call.
    • Determines users’ capabilities in terms of media and encoding types.
  • Session Setup: Establishes session setup by defining parameters (ports).
  • Session Management: Provides call holding, call forwarding, accepting new participants, and changing session parameters.

H.323 Protocols

H.323 is a complete set of protocols for establishing and maintaining voice (or video) communication, unlike SIP, which is only a signaling protocol. Key components include:

  • G.711 or G.723.1: Compression algorithms.
  • H.245: Allows parties to negotiate the compression method.
  • Q.931: Establishes and terminates connections.
  • H.225: Registration with the gatekeeper.
  • RTP and RTCP: H.323 mandates the use of these protocols.

SIP vs. H.323

Key differences between SIP and H.323:

  • H.323 defines a complete set of protocols for real-time interactive multimedia (signaling, registration, admission control, transport, and compression). SIP is primarily for initialization, management, and termination of sessions.
  • H.323 was defined by ITU-T (telephony), while SIP was proposed by IETF (Internet).
  • H.323 is a complex standard, while SIP is simpler and easier to implement.

Digital Watermarking

A watermark is created and inserted into the original digital content to produce watermarked content.

Watermark Embedding Techniques

Two main watermark embedding techniques exist:

  • Coefficient-Based Approaches: Embedding is performed by directly modifying pixel values.
  • System-Based Techniques: Embedding is performed by slightly changing an existing processing system.

Coefficient-Based Watermarking

Embedding is performed by directly modifying pixel values.

  • Simplest method: The watermark is added to the original media similarly to additive noise.
  • Other methods insert the watermark after the media has been transformed.

Watermark Insertion Methods

  • Spatial Domain: Watermark is inserted by operating on the pixels of the image or video.
  • Transformation Domain: Watermark is inserted after a transformation, including techniques like:
    • Discrete Fourier Transform (DFT)
    • Discrete Cosine Transform (DCT)
    • Wavelet Transform (WT)
    All of these are mathematical methods to transform from time-domain data to an equivalent representation in the frequency domain.