UNIT 3

The Network Layer

1
 Network Layer Design Issues
• Store-and-Forward Packet Switching
• Services Provided to the Transport Layer
• Implementation of Connectionless Service
• Implementation of Connection-Oriented Service
• Comparison of Virtual-Circuit and Datagram
Subnets

2
 Network Layer Design Issues
• Network layer provides point-to-point connectivity
between any two hosts.
• The network layer services have the following goals:
– The services should be independent of the router technology.
– The transport layer should be shielded from the number,
type, and topology of the routers present.
– The network addresses made available to the transport layer
should use a uniform numbering plan, even across LANS
and WANS.
• The network layer defines the service provided by the
subnet. A subnet (short for "subnetwork") is an
identifiably separate part of an organization's network.
3
Store-and-Forward Packet Switching

fig 5-1

The environment of the network layer protocols.

4
 Functions of Network Layer
• Routing – find a path from one host to another host.
• Congestion control – mechanisms to prevent hosts
from flooding the network.
• Quality of Service (QoS) - transmission rates, error
rates, and other characteristics can be measured,
improved, and, to some extent, guaranteed in advance.
• Internetworking provides translation between subnet
using different protocols.

5
Services Provided to Transport Layer
• The freedom in writing detailed specifications of the
services to be offered to the transport layer cause
battles between connection-oriented and
connectionless services.
• Internet community - connectionless
– With 30 year experience with the Internet, the subnet is
inherently unreliable.
– The host should accept this fact and do error control and
flow control themselves.
• Telephone companies – connection-oriented
– With more than 100 years’ experience, QoS is important.
• QoS is important and the Internet is starting to
associate with connection-oriented service. 6
 Implementation of Services
• Connectionless service
– No advance setup is needed.
– The packets are frequently called datagrams.
– The subnet is called a datagram subnet.
– The routing algorithm is the algorithm that manages the
tables and makes the routing decision.
• Connection-oriented service
– A path from the source router to the destination router must
be established before any data packets can be sent.
– The connection is called a VC (virtual circuit).
– The subnet is called a virtual-circuit subnet.
– To distinguish packets from different hosts, replacing
connection identifiers in outgoing packets is called label
switching. 7
Implementation of Connectionless Service

Routing within a diagram subnet.

8
Implementation of Connection-Oriented Service

Routing within a virtual-circuit subnet.

9
Comparison of Virtual-Circuit and
Datagram Subnets

5-4

10
 Routing Algorithms
• The Optimality Principle
• Shortest Path Routing
• Flooding
• Distance Vector Routing
• Link State Routing
• Hierarchical Routing
• Broadcast Routing
• Multicast Routing

11
 Routing Algorithms

• The routing algorithm is a part of network layer software to

decide which output line an incoming packet should be
transmitted on.
• Session routing is a route remains in force for an entire user
session.
• Routing algorithms should be correctness, simplicity,
robustness, stability, fairness, and optimality.

12
 Routing Algorithms
• Non-adaptive algorithms
– They do not base their routing decisions on
measurements or estimates of the current traffic and
topology.
– This procedure is sometimes called static routing.
• Adaptive algorithms
– They change their routing decisions to reflect
changes in the topology.
– This procedure is sometimes called dynamic
routing.
13
 The Optimality Principle
• If the router J is on the optimal path from the router I
to the router K, then the optimal path from J to K also
falls along the same route.
– Construct a sink tree with the destination to be root.
– The goal of all routing algorithms is to discover and use the
sink tree for all routers.
– Since it is a tree, there is no loops.
– A real network is complex. Routers and links may be down
at any time.

14
 Sink Tree

(a) A subnet. (b) A sink tree for router B.

15
 Shortest Path Routing
• Shortest Path Routing is a static routing algorithm that
just finds the shortest path.
• A graph is used to represent the network.
– Each node of the graph represents a router.
– Each arc of the graph represents a communication link.
– To choose the route between a given pair of routers, the
algorithm just finds the shortest path between them on the
graph.
• Metric used in the shortest path.
– Number of hops
– Geographic distance in miles/kilometers
– Transmission delay  fastest path
16
 Shortest Path Routing
• Dijkstra Algorithm
– Each arc (link) is labeled with a weight (link distance).
– Each node is labeled with the distance from the source node
along the best known path and the source node.
– Initially, no paths are known, all nodes except the source are
labeled as (∞, -).
– All labels may be either tentative or permanent. Initially, the
labels are tentative. When it is discovered to be shortest
possible path, the label is made permanent and never
changed thereafter..

17
 Shortest Path Routing
• An example: find the shortest path from A to D
– We start out by making node A permanent indicated
by a filled-in circle.
– Then we examine each node adjacent to A,
relabeling each one.
– Scan all the tentatively labeled nodes in the whole
graph and make the one with the smallest distance
to A permanent.
– This node becomes the new working node. Repeat
the steps till the destination becomes permanent.
18
Shortest Path Routing

19
 Flooding
• Flooding is a static routing algorithm.
• Every incoming packet is sent out on every outgoing
line except the one it arrived on.
• Flooding generates a large number of duplicated
packets. To reduce overhead,
– Use a hop counter (TTL, Time To Live), which is
decremented at each hop. The packet is discarded with the
counter reaches zero.
– Keep track of the packets and avoid to send them out the
second time in case there is a loop.
– Selective flooding in which the routers send the incoming
packet to only those outgoing lines in the right direction.
• Flooding has tremendous reliability and always choose
the shortest delay used in applications such as military,
distributed database, wireless network.
20
 Distance Vector Routing
• Distance Vector Routing algorithm – Dynamic routing
– Each router maintains a table (vector), giving the best known
distance to each destination and the outgoing line to get there.
– These tables are updated by exchanging information with the
neighbors.
– The metric used might be the number of hops, the time delay,
or the number of queued packets.
– The router is assumed to know the “distance” to each of its
neighbors.

21
 22
(a) A subnet. (b) Input from A, I, H, K, and the new routing table for J.
 Distance Vector Routing
• Distance vector works in theory but has a serious
drawback in practice.
– React rapidly to good news when a router comes up.
– Though it finally converge to correct result, it takes long time
when where is a bad news.
– There are several attempts to solve the problem, but none is
perfect.
• Distance vector routing was used in ARPANET until
1979 when it is replaced by link state routing.
• Two problems of distance vector routing:
– It does not take line bandwidth into account.
– It took too long to converge. 23
The count-to-infinity problem.

24
 Link State Routing
• Link State Routing is a dynamic routing.
• Each router must do the following:
1. Discover its neighbors, learn their network
address.
2. Measure the delay or cost to each of its neighbors.
3. Construct a packet telling all it has just learned.
4. Send this packet to all other routers.
5. Compute the shortest path to every other router.

25
 Learning about the Neighbors
• Learning about the neighbors: When a router is
booted, it first learns its immediate neighbors.
– Send a HELLO packet on each point-to-point line.
The router on the other end will send a reply telling
who it is.
– Each router has a global unique name.
– If two or more routers are connected by a LAN, we
can model the LAN as a node.

26
 Learning about the Neighbors

(a) Nine routers and a LAN. (b) A graph model of (a).

27
 Measuring Line Cost
• Measuring Line Cost
– Send an ECHO packet, measure the round trip
delay, and divide it by two.
• Repeat it several items to have a better
estimation.
– Whether to take the load into the account?
• Consider the load: start measuring delay when
ECHO is queued. Choosing unloaded line results
in better performance. But the load might
oscillate.
• Ignore the load: start measuring delay when the
ECHO packet reaches the front of the queue.
28
 Measuring Line Cost

A subnet in which the East and West parts are connected by two lines.
29
 Building Link State Packets
• Build the link state packet containing: node ID,
sequence number, age, a list of neighbors and the delay
to the neighbor.
• Building the state packet is easy. The hard part is to
determine when to build them.
– Periodically or event-driven

(a) A subnet. (b) The link state packets for this subnet. 30
 Distributing the Link State Packets
• The trickiest part is to distribute link state packet.
• Basic idea:
– Use flooding to distribute the link state packets.
– To keep the flood in check, each packet contains a
sequence number that is increased by one for each new
packet.
– When the link state packet arrives, the router check if
it is new.
• Yes  forward it to all outgoing lines except the one it
arrived.
• No (duplicated or with low sequence number)  discard it.
31
 Distributing the Link State Packets
• Potential problems:
– The sequence number wrap around  use the 32-bit
sequence number. It takes 137 years to wrap around.
– The router crashes. Its sequence number starts again
from 0, it is rejected.
– The sequence number is corrupt (e.g., 65540 is
received instead of 4, then packets from 5 to 65540
will be rejected.)
• Use “age” to solve the problems:
– The age decreases by one per second. The packet is
discarded when age = 0.
– Problem packets won’t last for a long time. 32
 Distributing the Link State Packets
• Each router uses a table to maintain the link state packets.
• Each row is a recently received but not processed packet.
• Each entry includes the source address, sequence number, age,
and send/ACK flags.

33
The packet buffer for router B in the previous slide (Fig. 5-13).
 Computing Routes
• Once a router has accumulated a full set of link state
packets, it knows all nodes and links, thus can construct
the subnet graph.
• Run Dijkstra algorithm to find the shortest paths from
the source to all other nodes.
• For a network with n routers, each with k neighbors, the
memory required in nk.
– Memory and computational time may be a problem for large
subnets.
– But it works fine for many practical situations.
• The OSPF (Open Shortest Path First) protocol is used in
the Internet.
• IS-IS (Intermediate System-Intermediate System) is used
in some the Internet backbone (NSFNET).
34
 Hierarchical Routing
• With the increase of network/routers, it is infeasible to
have an entry for each router. The hierarchical routing is
required.
– Divide the routers into regions.
– The router only knows details to route packets to the
destination within the same region.
– For Huge Networks a two level hierarchy may be insufficient
– It may be necessary to group Regions to Clusters, Clusters
to Zones and Zones into Groups.

35
Hierarchical Routing

36
 Broadcast Routing
• Broadcasting: send a packet to all destinations.
– Distributing weather reports, stock, radio programs, etc.
• Broadcast routing algorithm
– Send a distinct packet to each destination (waste bandwidth)
– Flooding (generate too many packets)
– Multi-destination routing
• The packet includes a list of destinations
• The router sends the packet on an outgoing line if it is the
best route for at least one of destinations (according to
routing table).

37
 Broadcast Routing
• Broadcast routing algorithm
– A spanning tree is a subset of the subnet that includes all the
routers but contains no loops.
• Copy an incoming broadcast packet onto all the spanning
tree lines except the one it arrived on.
• excellent use of bandwidth
• But each router is required to know some spanning tree.

38
 Multicast Routing
• Sending a packet to a group of nodes (a subset of the
nodes in the network) is called multicasting.
– Multiple unicast or broadcast are too expensive
– Build spanning tree
– Upon receiving a packet, prune the spanning tree (cut off the
routers/lines that do not lead to any member in the group)
– Not scalable

39
 Multicast Routing

(a) A network. (b) A spanning tree for the leftmost router. 40

(c) A multicast tree for group 1. (d) A multicast tree for group 2.
THANK YOU