STRUCTURE OF A SWITCH

We use switches in circuit-switched and packet-switched networks.

Structure of Circuit Switches
 Circuit switching can use either of two technologies:
1. Space-division switching or
2. Time-division switching

1. Space-Division Switching
o Here, paths in the circuit are separated from one another spatially.
o It was designed for analog networks.
o Currently it is used in both analog and digital networks.

1.1 Crossbar Switch

 It connects n inputs to m outputs in a grid, using electronic microswitches
(transistors) at each crosspoint.

Fig. Crossbar switch with three inputs and four outputs

 Limitation: More number of crosspoints required.
 To connect n inputs to m outputs using a crossbar switch requires n × m
crosspoints.
 Example: To connect 1000 inputs to 1000 outputs requires a switch with
1,000,000 crosspoints.
 A crossbar switch with this number of crosspoints is impractical.
1.2 Multistage Switch
 It is the solution to the limitations of the crossbar switch.
 It combines crossbar switches in several (normally three) stages.
 In a single crossbar switch, only one row or column (one path) is active for any
connection. So we need N × N (that is N2) crosspoints.
 If we allow multiple paths inside the switch, we can decrease number of crosspoints.
 Each crosspoint in the middle stage can be accessed by multiple crosspoints in the
first or third stage.

To design a three-stage switch, we follow these steps:

1. Divide the N input lines into groups.
 Each group has n lines.
 For each group, we use one crossbar of size n × k, where k is the number of
crossbars in the middle stage.
 In other words, the first stage has N/n crossbars of n × k crosspoints.
2. Use k crossbars, each of size (N/n) × (N/n) in the middle stage.
3. Use N/n crossbars, each of size k × n at the third stage.

 Calculate the total number of crosspoints in a three-stage switch is as follows:

 Total number of crosspoints = 2kN + k (N/n)2

which is much smaller than the number of crosspoints in a single-stage switch
(N2).
 Multistage switch has one drawback:
o Blocking during periods of heavy traffic.
 Blocking refers to times when one input cannot be connected to an
output because there is no path available between them—all the
possible intermediate switches are occupied.
o Sharing the crosspoints in the middle-stage crossbars.
 Sharing can cause a lack of availability if the resources are limited
and all users want a connection at the same time.
o As the number of stages increases, possible blocking increases as well.
 In single-stage switch, blocking does not occur because every combination of input
and output has its own crosspoint; there is always a path.

 Clos investigated the condition of nonblocking in multistage switches and came up

with the following formula.
 In a nonblocking switch, number of middle-stage switches must be at least 2n – 1.
 In other words, we need to have k ≥ 2n - 1.
According to Clos criterion: n = (N/2)1/2 and k ≥ 2n - 1
Total number of crosspoints ≥ 4N [(2N)1/2 - 1]
2. Time-Division Switching
 It uses time-division multiplexing (TDM) inside a switch.

2.1 Time-Slot Interchange (TSI)

 It is the most popular technology.

Figure shows a system connecting four input lines to four output lines.

 Imagine: each input line wants to send data to an output line according to the following
pattern:
(1→3), (2→ 4), (3 →1), and (4 → 2), in which arrow means “to”.
 Figure combines a TDM multiplexer, a TDM demultiplexer, and a TSI.
 TSI consisting of random access memory (RAM) with several memory locations.
 Size of each location is same as size of a single time slot.
 Number of locations is same as number of inputs.
 RAM fills up with incoming data from time slots in the order received.
 Slots are then sent out in an order based on the decisions of a control unit.
3. Time- and Space-Division Switch Combinations
 Space-division switching
o Advantage : it is instantaneous.
o Disadvantage: number of crosspoints required to make space-division
switching acceptable in terms of blocking.
 Time-division switching
o Advantage: it needs no crosspoints.
o Disadvantage: In TSI, processing of each connection creates delays. Each
time slot must be stored by RAM, then retrieved and passed on.

 Time- and Space-Division Switch

o We can combine both the technologies to take advantage of both.
o Multistage switches can be designed as time-space-time (TST) switches.

Figure shows a simple TST switch that consists of two time stages and one space
stage and has 12 inputs and 12 outputs.
 Instead of one time-division switch, it divides the inputs into three groups (of four
inputs each) and directs them to three timeslot interchanges.
 Here, average delay is one-third of one time-slot interchange to handle all 12 inputs.
 Last stage is a mirror image of first stage.
 Middle stage is a space-division switch (crossbar) that connects the TSI groups to
allow connectivity between all possible input and output pairs.
Structure of Packet Switches
 Packet switch has a different structure from circuit switch.
 Packet switch has four components:
1. Input ports
2. Output ports
3. Routing processor
4. Switching fabric

1. Input Ports
 It has physical-layer processor and a data-link processor.
 It performs physical and data-link functions of packet switch.
 The bits are constructed from received signal.
 Packet is decapsulated from frame.
 Errors are detected and corrected.
 Packet is now ready to be routed by network layer.
 It has buffers (queues) to hold the packet before it is directed to switching fabric.

Figure shows a schematic diagram of an input port

2. Output Port
 It performs same functions as input port, but in reverse order.
 First outgoing packets are queued.
 Then packet is encapsulated in a frame.
 Finally physical-layer functions are applied to the frame to create the signal.

Figure shows a schematic diagram of an output port.

3. Routing Processor
 It performs network layer functions.
 Destination address is used to find
o Address of next hop
o Output port number from which the packet is sent out.
 This activity is referred to as table lookup because the routing processor searches
the routing table.

4. Switching Fabrics
 The most difficult task is to move packet from input queue to output queue.
 Packet switches are specialized mechanisms that use a variety of switching fabrics.
a. Crossbar Switch
 It is the simplest type of switching fabric.
b. Banyan Switch
 It is a more realistic approach than crossbar switch.
 It is a multistage switch with microswitches at each stage that route the packets
based on output port represented as a binary string.
 For n inputs and n outputs, we have log2 n stages with n/2 microswitches at each
stage.
 First stage routes the packet based on the high-order bit of the binary string.
 Second stage routes the packet based on the second high-order bit, and so on.
 Number of stages is log2(8) = 3.

Figure shows a banyan switch with eight inputs and eight outputs
 A packet has arrived at input port 1 and must go to output port 6 (110 in binary).
 First microswitch (A-2) routes the packet based on first bit (1).
 Second microswitch (B-4) routes the packet based on second bit (1).
 Third microswitch (C-4) routes the packet based on third bit (0).

 A packet has arrived at input port 5 and must go to output port 2 (010 in binary).
 First microswitch (A-2) routes the packet based on first bit (0).
 Second microswitch (B-2) routes the packet based on second bit (1).
 Third microswitch (C-2) routes the packet based on third bit (0).
c. Batcher-Banyan Switch
 Problem with the banyan switch: possibility of internal collision even when two
packets are not heading for the same output port.
 We can solve this problem by sorting the arriving packets based on their
destination port.
 K. E. Batcher designed a switch that comes before the banyan switch and sorts
the incoming packets according to their final destinations.
 The combination is called the Batcher-banyan switch.

Fig. Batcher-banyan switch

 Hardware module called a trap is added between the Batcher switch and the
banyan switch.
 Trap module prevents duplicate packets (the packets with the same output
destination) from passing to the banyan switch simultaneously.
 Only one packet for each destination is allowed at each tick.
 If there is more than one, they wait for the next tick.