Things to check
• Branch IDs get generated in strict way in RFC3261.  Not currently reflected here.

Default ports

• TCP/UDP 5060 (RTP/RTCP anything UDP, often above 40000).

Components

• SIP Gateway - translates to another an outside network. considered a special type of User Agent
• User Agent
  

Cisco Notes

• UCM does not control SIP devices, and SIP devices do not register with UCM.
  

Reqs and Responses

Reqs

SIP Network Elements

• UAC (User agent client) - initiates SIP session
• UAS (User agent server) - Contacts user when SIP request is recieved and returns response on behalf of user.
• SIP Servers
  • Registrar server - Keeps track of cur location of UACs
  • Proxy server - recieves requests from a client and forward them on the client's behalf
  • Redirect server - provides client with next hop/hops a msg must take. Cilent can than directly contact the appropriate next hop.

Basic Sessions

Conference Call - SDP and SAP

• Use SDP and SAP.
• SDP describes multimedia sessions to announce session and invite endpoints.  
• SAP delivers SDP packets using multicasting where participants are not known in advance.  Uses SDP to describe the session being multicast;  anounces conf session by periodically multicasting announcement packet to well-known multicast addr and portr.

• UDP (default port 5060)
• TCP (default port 5060) - Extra cost and complexity
• TLS (default port 5061) - doesn't work well across proxy, use S/MIME instead
• SCTP

• RFC2327 - SDP
• RFC3261 - SIP: Session Initiation Protocol
• RFC3262 - Reliability of Provisional Responses in Session Initiation Protocol (SIP)
• RFC3263 -  Session Initiation Protocol (SIP): Locating SIP Servers 
• RFC3264 - An Offer/Answer Model with Session Description Protocol (SDP)
• RFC3265 -  Session Initiation Protocol (SIP)-Specific Event Notification
• RFC3266 -  Support for IPv6 in Session Description Protocol (SDP) 
• RFC3267 -  Real-Time Transport Protocol (RTP) Payload Format and File Storage Format for the Adaptive Multi-Rate (AMR) and Adaptive Multi-Rate Wideband (AMR-WB) Audio Codecs
• RFC3550 - RTP and RTCP (RTP Control Protocol)
• RFC3551 - RTP Audio/Video profiles.  (see below for summary list).
• RFC4566 - SDP

Notes

• DNS SRV queries used to locate proxy server in other domain
• Multicast IPv4 address is 224.0.1.75
• FWs - RTP can be on various ports, and is different from SIP session. Use ALG/B2BUAs to get through. UA goes to ALG. FW allows SIP and RTP traffic to/from the ALG, and blocks everything else.
• FWs - If FW is SIP aware, it can open up 'pinholes' dynamically by parsing SIP headers. But it has to work with inbound traffic also.
• STUN protocol (port 3478) allows client to discover presence and type of NATs between client and public Internet. 
  • Shared secret (usually a username/pw). 
  • TLS over TCP for initial STUN connection.  
  • Client then sends a Shared Secret Req packet to server and receives the shared secret to be used.
  • Sends binding req packets (UDP) to svr.  Response from svr tells client public IP addr * port #.
  • Multiple binding req packets can help determine type of NAT.

    Type	Translation
    Internet	No translation; client has a pubilc IP address
    Static NAT	one-to-one mapping (Juniper/Netscreen calls this MIP (mapped IP))
    (Address) Full cone NAT	Constant mapping between private IP addr and public ip addr many-to-one (hide nat) one-to-many (type of pat - Juniper/Netscreen calls this VIP (Virtual IP)) - IPAddrA/PortA -> IPAddrX/PortA IPAddrA/PortB -> IPAddrY/PortA IPAddrA/PortC -> IPAddrY/PortC pat can also have ranges...
    Dynamic Port NAT	External is one IP address, multiple ports.  Can be Full Cone, Restricted Cone, or Port Restricted Cone
    Dynamic Pooled NAT	External is pool of IP addresses.  Internal addresses assigned dynamically as required to external.  Can be Full Cone, Restricted Cone, Port Restricted Cone.
    Restricted cone NAT	Constant mapping, but an outgoing packet is needed to open incoming path.  May or may not include port in mapping.
    Port Restricted cone NAT	Same as Restricted cone NAT, but includes port in mapping.
    Symmetric NAT	Diff public IP addr mapping is used based on dest IP addr

• For Symmetric NAT, and certain topologies such as communication between 2 UAs behind NATs, signaling and media relay protocol such as TURN is required.
• TURN protocol - allows client to obtain transport IP addr and port that can receive packets sent from a single IP addr in public Internet.  TURN server becomes a relay.  Obviously can introduce packet latency, and increases probability of packet loss.  "TURN should only be used when it is the only approach available (i.e. IP address obtained using STUN will not work)" (from Alan Johnston book).
  • Reuses Shared Secret Request and Shared Secret Response packets to establish r  AlsoBshared secret.
  • Binding Request and Binding Response packets detect and categorize NAT in the path.
  • Allocate Request to request a relay IP address and port number be returned in an Allocate Response packet.
• Other solutions include using TCP & rport parameter in VIA.

*required

<---(RTP SESSION)--->

INVITE msg - Proxy(address-of-record)->ProxiedContact(device/endpoint URI)

INVITE sip:jungar@alpha-org.com SIP/2.0	*method, Request-URI, SIP version #
Via: SIP/2.0/UDP proxy.alpha-org.com:5060;branch=as82je8ei4kr Via: SIP/2.0/UDP 100.101.102.103:5060	*proxy VIA pre-pends original VIA 100.101.102.103 is numeric addr of end device that invite came from SIP version #/transport hostname-or-addr:portnum;branchparam
Max-Forwards: 69	decremented as it went through proxy
To: J. Ungar <sip:jungar@alpha-org.com>	*shows destination of SIP request.
From: J. Smith <sip:jsmith@beta-org.com>;tag=13456	*shows origination of SIP request.  Tag is combined with Call-ID to create unique conversation.
Call-ID:123456789@100.101.102.103	*keeps track of particular SIP session.
CSeq: 1 INVITE	*Starting integer (incremented with each new request sent), method
Subject: Let's talk	Optional.  Could be displayed during alerting.
Contact: <sip:jsmith@zetamachine.beta-org.com>	SIP URI of J. Smith's communication device.
Content-Type: application/sdp	message body is SDP
Content-Length: 158	message body has 158 bytes (including CR/LFs)
v=0	SDP protocol version #
o=Smith 2890844526 2890844526 IN IP4 100.101.102.103	Originator info: username, sess-ID ver net-type addr-type address
s=Phone Call	name of session
c=IN IP4 100.101.102.103	Connection Data-Addr that will be sending media packets: network-type address-type connection-address
t=0 0	start/stop time of session.
m=audio 49170 RTP/AVP 0	Media info: media port transport format-list
a=rtpmap:0 PCMU/8000	attributes of preceding session (RTP/AVP or UDP)

*required

180 Ringing ProxiedContact->Proxy

SIP/20.0 180 Ringing	copied from invite
Via:SIP/2.0/UDPproxy.alpha-org.com:5060;branch=as82je8ei4kr;received=100.101.102.105 Via: SIP/2.0/UDP 100.101.102.103:5060	copied from invite; recived parameter contains literal IP addr that req was recieved from; First VIA will be stripped by Proxy on way to smith
To: J. Ungar <sip:jungar@alpha-org.com>;tag=643a33	copied from invite (same order);tag added
From: J. Smith <sip:jsmith@beta-org.com>;tag=13456	copied from invite (same order)
Call-ID:123456789@100.101.102.103	copied from invite
CSeq: 1 INVITE	copied from invite
Contact: <sip:jungar@200.201.202.203>	contact where to can be contacted once session is established
Content-Length:0	no content...

200 Ringing ProxiedContact->Proxy

SIP/20.0 180 OK	copied from invite
Via:SIP/2.0/UDPproxy.alpha-org.com:5060;branch=as82je8ei4kr;received=100.101.102.105 Via: SIP/2.0/UDP 100.101.102.103:5060	copied from invite; recived parameter contains literal IP addr that req was recieved from; First VIA will be stripped by Proxy on way to smith
To: J. Ungar <sip:jungar@alpha-org.com>;tag=643a33	copied from invite (same order);tag added
From: J. Smith <sip:jsmith@beta-org.com>;tag=13456	copied from invite (same order)
Call-ID:123456789@100.101.102.103	copied from invite
CSeq: 1 INVITE	copied from invite
Contact: <sip:jungar@200.201.202.203>	contact where to can be contacted once session is established
Content-Length:0	no content...
v=0
o=ungar 2890844528 2890844528 IN IP4 200.201.202.203
s=Phone Call
c=IN IP4 200.201.202.203
t=0 0
m=audio 49172 RTP/AVP 0
a=rtpmap:0 PCMU/8000

RTP packet header

 0                   1                   2                   3
 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|V=2|P|X|  CC   |M|     PT      |        sequence number        |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|                            timestamp                          |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|            synchronization source (SSRC) identifier           |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|             contributing source (CSRC) identifiers            |
|                               ....                            |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

• V - version
• P - Padding. If this bit is set, padding octets are added to end of packet to make it fixed length. Used for encryption.
• X - Extension.  If set, there is one additional extension following header.
• CC - CSRC count.  number of content source identifiers present following header.  Used by emixers that take multiple RTP streams and output a single stream.
• M - Marker.  Indicates start of new frame in video, or start of talk-spurt in silence-suppresed speach.
• PT - Payload Type.  defines codec in use.  Should match profile # listed in SDP.
• Sequence number.  Incremented for each RTP packet sent.  Used to detect missing/out of sequence packets.
• Timestamp.  indicates in relative terms the time when payload was sampled.  Allows receiver to remove jitter and play back packets at right interval assuming sufficient buffering.
• SSRC (Sync Src ID) - randomly generated ID that client uses to id session. IDs sender of RTP packet.  aAt start of session, each participant chooses a SSRC number randomly.  If 2 participants choose the same number, they each choose again until each party is unique.
• CSRC Contributing Source ID - 0-15 instances (based on CC).  Can be used for a mixer. Preserves ID of original src.