Overview: IP

• IPv6
• mobile IP
• ICMP
• DHCP
• SLIP
• PPP
• IP routing

IPv6

• design initiated in response to exhaustion of class B addresses
• 128-bit addresses
• better autoconfiguration
• better security (encryption, authentication)
• better type of service
• simpler to process a packet
• no header checksum
• "extension headers" instead of options, protocol field

ICMP

• Internet Control Message Protocol
• ping
• try to inform sender when discarding a packet
• (never send in response to an ICMP packet)
• unreliable, but useful

Mobile IP

• post office forwarding address
• I get a guest IP address from the network I'm on
• forwarding agent on my network sends all my packets to the guest address
• I can either
  • send packets with my home IP address (probably dropped for security reasons)
  • send packets with my guest IP address (source address authentication won't work)

DHCP

• Dynamic Host Configuration Protocol
• "plug and play"
• broadcast request for IP address
• assigned IP address
• e.g. ISPs

PPP

• "lower layer" for IP (and other protocols)
• has IP address negotiation
• can use multiple serial lines
• 2-byte CRC
• complex, useful

SLIP

• "lower layer" for IP only
• marker character END, replaced in packet by ESC END
• escape character ESC, replaced in packet by ESC ESC
• END starts and ends packet
• IP+TCP headers are predictable: compression increases rate of useful data

Routing

• find a "good" path from one router to all other routers
• "good" is abstracted as "minimum metric"
• metric can be cost, distance, or simply 1 for a functioning link, infinite for a non-functional link
• in practice, low cost links can be short, or high bandwidth, or underutilized

Dijkstra's shortest path algorithm

• given:
  • a set of nodes n in N
  • a starting node a in N (N' = N - {a}
  • a set of links l in N * N
  • link metrics m(l) > 0
• calculates distance d(n) from a to each node in N'

Dijkstra's shortest path algorithm

1. forall n in (N - {A}) set d(n) = oo
2. the set of nodes with known distances is K: initially, K = {}.
3. loop: compute the distance for each node in n in N: d(n) = min(d(n), min[n' in (N - n), l in (n' * n)](d(n')+m(l))
4. find the node n in (N' - K) with minimum distance d(n), add the node to K
5. if N' = K, we are done, otherwise return to step "loop"

why algorithm works

• node distances (d(n)) can only decrease, never increase
• if a path to n goes through n', then at the end d(n) > d(n'), since forall l, m(l) > 0
• if a node is chosen for K, its distance is less than or equal to distance of all nodes still in N - K
• therefore, the distance of all nodes in K is less than or equal to the distance of each node in N - K
• therefore, the distance of all nodes in K can never decrease beyond the current value of d(n)
• therefore, the distance of all nodes in K is the shortest path distance