P a g e |1
About Me
I am shashank singh. A computer science and engineering grad from IET (Govt.)
Lucknow
. While I was preparing for competitive exams, I made a collection of formulafrom
various sources. Then I thought that it would be great to share it with my fellow
colleagues who are also on the same boat. This inspired me to write this book so that
any GATE, PSUs and other aspirants in computer science could get all the tricky
formulas in one place.
P.S. This book should be used as a quick reference guide only for formulas. For concepts
please go through some standard text books. Also depending on your feedback, I am
planning to write a new edition of this book with more formulas and numerical problems
in it. Also please provide your valuable feedback as I am planning to publish another
quick peek reference book on important facts of computer science subjects so that
you can revise your subjects quickly.
For any feedback or suggestions please follow me on Facebook:
https://www.facebook.com/Shashank.bluee
Cuteiipie.blogspot.com
Thanks and Best wishes for your exams
P a g e |2
COMPUTER NETWORKS
Delay Calculations:
Propagation delay =
Transmission delay =
Channel Utilization:
1. For Ethernet, channel utilization, u
, where a =
1.1. for 1- persistent CSMA/CD LAN 10-1000Mbps speed, simplified channel efficiency is,
, where A =
probability p to transmit in a contention slot.
, k stations each with
is the time to transmit a frame.
is the worst case
one way propagation delay.
1.2. Efficiency of 802.3 Ethernet at 10Mbps with 512 bit connection:
, where, B = network bandwidth. L = Cable Length. C
= speed of signal propagation. E = optimal number if contention slots per frame. F = frame size
2. For Token ring (release after transmission or early token release),
2.1. Channel utilization, u
, where N is the number of stations in the ring.
3. For Token ring (release after reception or delayed token release),
3.1. Channel utilization, u
, where N is the number of stations in the ring.
4. For unrestricted simplex protocol: If a frame has d data bits and h overhead bits and channel
bandwidth = b bits/sec then,
4.1. Maximum channel utilization =
4.2. Maximum data throughput =
5. For stop-and-wait,
5.1. Channel utilization,
u=
, where a =
and p is the probability that a
frame is in error.
5.2. Also Maximum channel utilization =
P a g e |3
5.3. Maximum data throughput
6. For Simplex positive acknowledgement with retransmission (PAR) protocol:
6.1. Maximum channel utilization and throughput is similar to stop-and-wait protocol when the effect
of errors is ignored.
7. For Sliding Window Protocols with window size of w,
7.1. Go-Back-N,
7.1.1.Channel utilization,
={
7.2. Selective reject,
7.2.1.Channel utilization
,u ={
7.3. Condition for maximum utilization or throughput is:
Throughput Calculations:
Throughput = Channel Utilization
Channel Bandwidth
Signal and Noise Calculations:
1. Signal to Noise Ratio (in decibels, dB) =
a. where S= Signal strength and N = noise strength.
2. Signal Attenuation (in decibels, dB) =
Data Rate and Channel Capacity Calculations:
1. Nyquist Theorem: Maximum data rate = 2H
and V is number of levels.
2. Shannons theorem: Channel capacity =
(Hz). (Note: here
bits/sec, where H is bandwidth in hertz (Hz)
(
) bits/sec, where H is bandwidth in hertz
is not in decibels).
Baud rate: A baud is the number of changes per second in the signal.
For Manchester encoding, baud rate = 2
MAC Sub layer:
Static channel allocation in LANs and MANs.
If C = channel capacity in bps
= arrival rate of frames (frames/sec)
= no. of bits per frame, then
bit-rate
P a g e |4
Mean time to delay, T =
Now, if the channel is divided into N sub channels each with capacity
each of the N channels is
T(fdm) =
and arrival rate or input rate on
then,
Dynamic Channel Allocation:
, where
is the arrival rate of frames.
Multiple access protocol:
Pure ALLOHA protocol
Infinite senders assumed.
Poisson distribution with mean rate S frames per frame time
Combined frame rate with retransmission G frames per frame time.
t is the time required to transmit a frame.
In multiple access protocol, a frame is successful if no other frames are transmitted in the
vulnerable period.
Probability of k frames being generated during a frame transmission time:
Hence, probability of zero frames in 2 frame periods is,
Therefore, for pure ALLOHA,
- Mean rate
which becomes maximum at G = ,
Max(S) =
= 0.184 = 18.4% throughput.
Vulnerability period in pure ALLOHA: For successful frame transmission, no other frame should
be on the channel for vulnerability period equal to twice the time to transmit one frame. That is,
, where t is the time to transmit one frame.
Slotted ALLOHA protocol
Time is divided into discrete frame slots.
A station is required to wait for the beginning of the next slot to transmit a frame.
Vulnerability period is halved as opposed to pure ALLOHA protocol. That is,
, where t is the time to transmit one frame.
Probability of k frames being generated during a frame transmission time:
P a g e |5
Hence, probability of zero frames in 1 frame period is,
- Mean rate
which becomes maximum at G = 1,
Max(S) =
PPP
= 0.368 = 36.8% throughput.
Expected number of retransmission,
In Point to Point Protocol (PPP), number of channels grows as square of the number of
computers. That is, (
Binary Exponential Backoff Algorithm:
After i collisions wait a random number of slots between 0 and
(Note: After 16 collisions, failure is reported to the higher layers.)
with a maximum of 1023.
Minimum frame size for IEEE 802.3 (Ethernet) frame = 64 bytes.
ROUTING ALGORITHMS
Shortest Past algorithm
Flooding Algorithm
Flow Based Routing Algorithm,
It uses the formula,
STATE
Non-Adaptive (or Static)
Non-Adaptive (or Static)
Delay time, T =
Non-Adaptive (or Static)
, where C= channel capacity,
1/ = mean packet size in bits, and = mean
number of arrivals in packets/sec
Distance Vector Routing (DVR)
Based on Bellman-Ford Algorithm and the
Ford-Fulkerson Algorithm.
It suffers from the Count to infinity
problem
Exchange information of the entire network
with its neighbors.
Link State Routing Algorithm (LSR)
Discovers its neighbors and construct a
packet telling all it has just learned and
send this packet to all other routers.
Hierarchical Routing Algorithm
For N router subnet, the total number of
levels =
And each router will have
number
of entries in their routing tables.
Broadcast Routing Algorithm
Adaptive Algorithm (or non-static)
Adaptive algorithm (or non-static)
Adaptive Algorithm (or non-static)
Adaptive Algorithm (or non-static)
Congestions deals with wires and routers while flow deals with hosts.
Traffic Shaping:
Leaky Bucket Algorithm (If the bucket overflows, then packets are discarded).
Token Bucket Algorithm (causes a token to be generated periodically).
Congestion control through Load Shedding may lead to deadlock and unfairness.
P a g e |6
The size of data portion of the datagram is = [Total length] [size of header]
Maximum length of the datagram = 65535 bytes.
Datagram format:
Total length
Header length
Flag length
Type of service
Identification bits
Fragment offset
Time to Live
Protocol version
Header Checksum
Source Address
Destination Address
In Layer 2 of OSI model (Data link layer), destination field appears before source field where as in
layer 3 (Network layer), the ordering is opposite.
IP class addressing:
Class Name
Class A
Class B
Class C
Class D
Class E
16 bits
4 bits
3 bits
8 bits
16 bits
13 bits
8 bits
8 bits
16 bits
32 bits
32 bits
Starts with
0
10
110
1110
11110
Range
0-127
128-191
192-223
224-239
240-255
Internet addresses refer to network connections rather than hosts. (For example, Gateways have two
or more network connections and each interface has its own IP address). Thus, Physical (or Ethernet)
address is constant (fixed) but Network (or IP address) may change.
Transport Layer:
To cope with the widely varying delays, TCP maintains a dynamic estimate of the current RTT (Round
Trip Time) calculated this way:
When sending a segment, the sender starts a timer.
Upon receipt of an acknowledgement, it stops the timer and record the actual elapsed delay M
between sending the segment and receiving its acknowledgement.
Whenever a new value M for the current RTT is measured, it is averaged into a smoothed RTT
depending on the last measured value as follows:
New RTT = a (Old RTT) + (1-a)(Current RTT M), where a is known as the smoothing factor and
it determines how much weight the new measurement carries. When a is 0, we simply use the
new value and when a is 1 we ignore the new value. Typically, the value of a lies between 0.8
and 0.9
TCP can provide reliable service where UDP cannot as they choose to use. IP in two different modes
of service provided by IP either reliable or connectionless.
P a g e |7
A transparent router connects a LAN to a WAN.
TCP does not have in-built mechanism to distinguish between a timeout owing to error and that owing
to congestion.
During early start phase, the congestion window inside TCP grows Exponentially, whereas after that it
grows linearly.
If two client try to connect to the same TCP service, then they can
Private network address enables reuse of IP addresses, thereby aiming to reduce the IP shortage
problem.
CLIENT SERVER MODEL:
CLIENT
SERVER
Socket
Socket
Bind
Connect
Listen()
Send()
Accept()
Receive()
Receive()
Close()
Send()
Terms connected with RPCs(Remote Procedural Calls):
Marshalling, Shunt
Client, Skelton
Procedure call, server
Terms not-connected with RPCs:
Socket, connect.
While trying to access a webpage www.facebook.com, a user gets the http error message could
not resolve host name. What could be the cause of this problem?
The local DNS is not running.
In TCP SYN flag is used for connection establishment.
A wants to send a group message to a number of people. How can it be ensured that the message
actually came from A and not from someone else?
A encrypts via its own private key; and the group decrypts via As public key.
The number of bits used for IPV6 addressing is 128 bits whereas for Ethernet address 48 bits are
used.
If n is the number of bits used to represent the frame sequence number then:
Name of sliding window
protocol
Sender window size
Receiver window size
1
Go-back-N
P a g e |8
Selective Repeat
The maximum burst rate at which network can handle the traffic is given by,
, where C = Capacity of bucket(in bits); M = Network rate(e.g., token ring with 10 Mbps
network LAN); R = arrival rate(e.g., rate of entering into the bucket).
In any sliding window protocol,
Works on IP Layer / Network layer.
Devices and their OSI Layer:
GATEWAY
Application Layer, Presentation Layer, Session
Layer, Transport Layer
Router
Network Layer
Bridges and Switches
Data Link Layer
Repeaters, Hubs, Amplifiers, Multiplexers
Physical Layer
A group of (
) routers are interconnected in a centralized binary tree, with a router at each
node. Router I communicate with router J by sending a message to the root of the tree. The root then
sends the message back down to J, then
P a g e |9
COMPUTER ORGANIZATION AND ARCHITECTURE
Interrupts
External
Interrupts
Maskable
Interrupts
Internal
Interrupts
Software
Interrupts
Non-Maskable
Interrupts
Maskable interrupts are enabled and disabled by executing instructions such as EI or DI. If the
computers interrupts is disabled, the computer ignores the maskable interrupt.
Interrupt is disabled -> Interrupt flag bit = 1
Interrupt is enabled -> Interrupt flag bit = 0
The Non-Maskable interrupt has higher priority than the maskable interrupt.
If both maskable and non-maskable interrupts are activated at the same time, the processor will
service the non-maskable interrupt first.
Non- maskable interrupts are typically used in power failure interrupts.
PIPELINE
(
Latch delay.
, where,
= pipeline segment delay; L =
P a g e | 10
Hazards of Pipeline:
Branch Instruction: For each branch instruction additional (n-1) pipeline clocks are required;
where n is the number of segments in the pipeline.
Data dependency
Resource Conflicts.
, where S=
no. of instructions; n = no. of pipe segments; C= Probability for an instruction to be a conditional
branch.
Cache works on the principle of locality.
Each process has its own virtual address space.
The unit block of data for transferring between disk and memory is called a page. (Typically 4kB)
TLB is actually a cache for the page table.
Virtual memory uses write back method because frequent access to disk is very slow.
A DRAM consists of 1 transistor and 1 capacitor.
An interrupt in which device supplies the interrupt requests as well as its address is called Vectored
Interrupt.
When INTER is encountered, the processor branches to the memory location which is
Determined by the RST N instruction given by the IO device.
Determined by the Call address instruction given by the IO device.
In 8085 microprocessor,
XCHG: Exchange the content of the HL and DE register pair respectively.
DAD H: Add HL pair with HL pair and store the result in HL pair.
Thrashing implies excessive page IO
In 8085, a stack pointer is a 16-bit register that points to stack.
, where Ic= number of
instructions; CPI=CPU clock cycles;
PARAM is a parallel computer.
ALE is a pin of an 8085 microprocessor that is used to the latch the 8 bits of address lines AD7-AD0
into an external latch.
The magnetic bubble memory has an access time of 30
The magnetic material that is normally used in bubble memories is GARNET.
P a g e | 11
For RELATIVE ADDRESSING:
Effective address = Program counter + Address part of instruction.
For example: 2 word instruction, I = JMP 65 stored in 2000 in decimal.
So EA = PC + Address part; EA = 2002 + 65 =2067 in decimal.
;e.g., 16M * 8 = 128 Mb
Memory Chip notation:
Chip notation:
For memory access:
Cycle Time = Latency Time + Transfer Time.
Maximum Forbidden latency:
A SIMD (Single instruction multiple data) is characterized by,
How many bits will be required to store a decimal number containing (a) 3 digit (b) 4 digit (c) 96 digits
; where n = number of columns in the reservation table.
notation.
)
)
)
Key formula is
(No. of possibilities in base 2) = (No. of possibilities in base 10) = (No.
of possibilities in any base)
Let N be the required number of bits and n be the number of decimal digits given, then
For page table:
)
)
; so for 96 decimal digit number we would need
. in binary number system
P a g e | 12
, P is the proportion of a program
that can be made parallel (i.e., benefit from parallelization); (1 P) is the proportion that cannot
be parallelized (remains serial).
P a g e | 13
THEORY OF COMPUTATION
If an NFA contains N no. of States then the corresponding DFA will contain maximum of
Identities for regular expressions:
states.
; Where P and Q are regular expression.
Ardens theorem:
If P & Q are two regular expressions over an alphabet S such that P does not contain
following equation:
in R has a unique solution(only one solution) i.e.,
Some important results:
If L1 is DCFL and L2 is regular, then L1 L2 is also DCFL
If L1 is DCFL and L2 is regular, then L1 L2 is also DCFL
Every regular language is DCFL
The union of DCFL & regular language is DCFL
All RE languages are TM enumerable.
The power set of an infinite set is not countable.
then the
P a g e | 14
Not all languages are recursively enumerable.
Recursive languages are closed under complementation but recursively enumerable languages
are NOT closed under complementation.
The complement of a CFL may or may not be closed under complementation.
The minimum number of states in DFA accepting strings containing number of 0s divisible by P and
1s divisible by Q =
The number of possible DFAs with N states and M input symbol =
For the strings over some alphabet
, the following are primitive regular expressions:
X, for each
Thus if | |
Every primitive regular expression is a regular expression.
then there are
primitive regular expressions defined over
For a set A of n distinct elements:
Number of symmetric and reflexive relations =
Number of symmetric relations =
Number of reflexive relations =
Number of irreflexive relations =
Number of transitive relations = 1,2,13,171,3994 for 1 ,2 ,3 ,4 , 5 ,.. elements
Number of anti-symmetric relations =
Number of reflexive and anti-symmetric relations =
Number of equivalence relations = 1,2,5,15,52,203 for 1,2,3,4,5,6, elements. (i.e., the BELL
number).
Number of all possible relations =
P a g e | 15
DATA STRUCTURE AND ALGORITHMS
C operators in order of precedence (highest to lowest). Their associativity indicates in what order
operators of equal precedence in an expression are applied.
Operator
()
Parentheses(function call)
[]
Brackets (array subscript)
Member selection via objet name
->
Member selection via pointer
++ --
Postfix increment / decrement
++ --
Prefix increment / decrement
+-
Unary Plus / minus
!~
Logial negation/bitwise complement
(type)
Associativity
Left-to-Right
Right-to-Left
Cast (convert value to temporary value of type)
Dereference
&
Address (of operand)
sizeof
Determine size in bytes on this implementation
*/%
Multiplication / division / modulus
Left-to-Right
Addition / Subtraction
Left-to-Right
Bitwise shift left, Bitwise shift right
Left-to-right
< <=
Relational less than/ less than or equal to
Left to right
> >=
Relational greater than/ greater than or equal to
+<< >>
== !==
Relational is equal to/is not equal to
Left to right
&
Bitwise AND
Left to right
Bitwise exclusive OR
Left to right
Bitwise inclusive OR
Left to right
Logical AND
Left to right
||
Logical OR
Left to right
?:
Ternary conditional
Right to left
Assignment
Right of left
&&
+= -=
Addition/subtraction assignment
*= /=
Multiplication / division assignment
%= &=
Modulus / bitwise AND assignment
^= |=
<<= >>=
,
Description
Bitwise exclusive / inclusive OR assignment
Bitwise shift left / right assignment
Comma (separate expression)
Radix sort is a non-comparative integer sorting algorithm.
Left to right
P a g e | 16
Worst case time complexity = O(kN); N number of numbers each with k digits.
Worst case space complexity = O( k + N )
Insertion sort is stable and inplace sorting.
Best case time complexity = O(n) and
worst case time complexity = O(
Binary Insertion sort
Best case time complexity = O(
Worst case time complexity = O(
)
)
For a look up in a B+ tree, we traverse from root node to leaf node. If there are K search keys and
number of pointers per node (or degree) is P, then for a look up,
Number of articulation points in a tree = number of internal nodes.
For a full K-ary tree,
Formula that computes the address LOC(A[J,K]) of A[J,K]
Column-major order:
Row major order:
[
[
where Array size = (M,N); M= number of rows; N= number of columns;
Relationships and definitions:
Asymptotic form
o(gn)
is the empty set. (
Relationship
Definition
P a g e | 17
For any real constant a and b, where b>0
Complexity ordering:
(
So space requirement for an inplace sort is O(1).
Bubble sort:
We need to make a maximum of (n-1) passes for n-input elements.
The average number of iterations of outer loop = O(n)
Total number of comparisons required =
Total number of sqaping required for arranging n elements in the sorted order by using bubble
sort =
And hence the average time complexity = O(
Space complexity of bubble sort = O(1)
The number of interchanges an never be greater than the number of comparisons.
If the file is already sorted then the time taken by bubble sort is order of O(n), as it will require
only one pass of (n-1) comparisons.
Quick sort:
Best and average case time complexity = O(n logn)
Worst case time complexity = O(
Worst case space complexity = O(n)
Insertion Sort:
Best case:
Time complexity = O(n)
Average and Worst case:
Time complexity = O(
Total number of comparisons are =
Selection Sort:
Best, average, and worst case time complexity = O(
Worst case space complexity = O(n) {total} and O(1) {auxiliary}
P a g e | 18
SHELL sort:
It is also known as diminishing increment sort.
Time complexity = O(
If n is an exact power of 2, then average running time is O(
Time Complexity = O(n logn)
Space complexity of Merge sort = O(
Binary Search:
Maximum number of comparisons is approximately
Interpolation search:
Time complexity = log (log n)
Worst case behaves like linear search = O(n)
Robust interpolation search worst complexity =
Hashing:
If n: no. of elements; m: no. of slots, and
Resolve by chaining unsuccessful search takes
Successful search takes
Double hashing:
Merge Sort:
Selection sort has the minimum number of swaps.
Here we use two different hash functions as
h(k,i) = (h1(k) + ih2(k)) mod m
Initial probe is to position T(h1(k))
Here
probe sequences are used.
In an open addressing scheme, the expected number of probes in an unsuccessful search is at most
assuming uniform hashing.
And expected number of probes in a successful search is at most
The total number of nodes in a complete binary tree =
Number of leaf nodes =
Number of non-leaf nodes =
Time complexity of matrix chain multiplication = O
Balance tree:
AVL tree
; where d is the depth.
P a g e | 19
Kruskals Algorithm:
The total time taken by this algorithm to find the minimum spanning tree is O
But the time complexity, if edges are not sorted is O
Where E: number of edges; V: number of vertices.
Prims Algorithm:
B-Tree.
The time complexity for Prims algorithm is O
The depth of a complete binary tree with n nodes:
The worst case height of AVL tree with n nodes =
Number of trees (not only binary trees) possible with n- nodes =
For a complete n-ary tree having either 0 or 1 sons. If k is the number of internal nodes of a complete
n-ary tree, then the number of leaves in it is given by =
Heapify() call takes O
Heap Sort:
The running time is O
A tighter bound is
times.
( )
Number of different unlabelled binary trees possible with n-nodes =
Number of different labeled binary tree possible with n-nodes =
The height of a red black tree with n- internal nodes is =
Suppose we use a hash function h to hash n distinct keys into a array T of length m. Assuming simple
( )
uniform hashing:
The expected number of collisions =
Number of moves required to solve the Tower of Hannoi Problem =
; where n is the number
of disks.
Dijkstras algorithm:
Bellman Ford Algorithm:
With V vertices and E edges, Time complexity =
With V vertices and E edges; Time complexity = O(EV)
Let T(l) and T(r) denote the left and right subtrees of a binary tree T. If T has n nodes, then the
Total Path Length, P(T) = P( l )+P( r ) + n-1 ; where P(l) is the path length of left subtree and P(r)
is the path length of right subtree.
P a g e | 20
Different implementation of Hashing and their time complexity:
Implementation
Insert
Search
Direct addressing
O(1)
O(1)
Ordered addressing
O(N)
O(log N)
Ordered List
O(N)
O(N)
Unordered array
O(1)
O(N)
Unordered List
O(1)
O(N)
Chromatic number of a bipartite graph = 2
Chromatic number of a planar graph
We can find the coloring of n vertices in O(
Time complexity of DFS with V vertices and E edges = O( V + E )
Time complexity of BFS with V vertices and E edges = O( V + E )
Total number of BST(binary search trees) with n distinct keys =
In complete k-ary tree, every internal node has exactly k children. The number of leaves in such a
4
)
( )
tree with n internal nodes is = n( k 1 ) + 1
For a hash table of size N,
Probability of inserting K integers with no collision =
Thus probability of collision for inserting K integers =
( )
Total number of stack permutations of a n distinct element is =
Total number of invalid permutations =
Time complexity of Bankers algorithm = O (
); where m= number of resources and n= number of
processes.
An n-dimensional hypercube (
) is simple undirected graph with
bit-strings of length n. Two vertices of (
vertices. The vertices of (
) are
) are connected by an edge only if u & v differs in exactly
one bit position.
Degree of each vertex =2 => is bipartite.
Number of edges =
) is Regular bipartite graph.
Floyd Warshall algorithm has time complexity = O(
Fibonacci series recurrence relation regardless of any initial condition is
P a g e | 21
T(n) = T(n-1) + T(n-2) + n =
Best case =
Worst case =
Fibonacci series recurrence relation regardless of any initial condition is
T(n) = T(n-1) + T(n-2) + K (constant)
Time complexity in all cases =
MASTER METHOD
It is applicable to only limited classes of recurrences which are in the form of,
( )
; and a >=1 and b >1 and fn is asymptotically positive.{that is, fn>0
for n>=n0}
Compare fn with
Case 1: If fn =
for some
Tn =
Case 2: If fn =
) for some
Tn =
Case 3: If fn =
) for some
and
Tn =
For example, Find the time complexity of the following recurrence relation:
( )
; so fn =nlogn =
Hence by case 2 of master theorem,
Tn =
Finding complexity by Substitution method for the following recurrence relation:
( )
Let n =
So T(
) = T(
Let T(
) = S(m)
) + C1
S(m) = S( ) + C1 = O(log m) = O(log log n)
P a g e | 22
DATABASE MANAGEMENT SYSTEM
Cross Product:
If n1 tuples in R and if n2 tuples in S
Then, in
,(
) tuples will be present
Performance measures on disks:
Access time: The time it takes from when a read or write request is issued to when data transfer
begins. It consists of:
SEEK TIME: The time it takes to reposition the arm over the correct track.
ROTATIONAL LATENCY: The time it takes to reposition the arm over the correct track.
Lock modes:
Read lock shared lock (s); write lock Exclusive lock (X)
R
OK
WAIT
WAIT
WAIT
P a g e | 23
OPERATING SYSTEM
Next CPU burst predicted is given by:
and tn is current CPU burst, T(n-1) is the past
predicted value, and Tn is the new predicted value.
For n fork() calls
Round Robin Algorithm:
child processes will be created.
If there are n processes in ready queue and time quantum is q, then each process gets
of the
CPU time in chunks of at most q time units.
Each process must wait no longer than (n-1)q time units until next time quantum.
Turnaround time also depends on the size of the time quantum.
If waiting time or fraction of each process is P and n is the number of processes, then
CPU utilization =
And probability that N processes will all wait at the same time =
Unix Inode:
Suppose there are 12 direct blocks and 1 indirect block and 1 double indirect block and 1 triple
indirect block then the maximum size of process supported by inode is
[
size).
; where BS = block size; ES = entry size (or block pointer
P a g e | 24
DIGITAL ELECTRONICS
Rs complement:
N: a +ve number in base r with an integer part of n digits, then
() > NOT > AND > OR
Total delay = (2N + 1)tp ; where tp= propagation delay of each gate.
N bit Look ahead Addrer:
Total delay = 4tp ; where tp= propagation delay of each gate.
If the n-bit decoded information has unused or dont care combinations, the decoded output will have
less than
N bit Binary parallel adder:
(R-1)s complement of N =
Operator Precedence:
0 and 0 for N = 0;
N: a +ve number in base r with an integer part of n digits and a fraction part of m digits, then
for N
R-1 complement:
Rs complement of N =
outputs.
The decoders is also called n-to-m line decoder where
Sequential Circuits:
With N-flip flops,
Ring counter can have N modulus counter
Johnson counter can have 2N modulus counter
Integer representation:
Signed Magnitude representation:
1s complement representation:
2s complement representation:
Duty cycle: A duty cycle is the percentage of one-period in which a signal is active.
So duty cycle,
P a g e | 25
SOFTWARE ENGINEERING
Risk estimation method:
Priority of each risk can be computed as:
; R= the likelihood of a risk coming true; S = the consequences of the problems
associated with that risk.
Basic COCOMO model:
Efforts in person-months,
Development time in months in person-months,
where a,b,c,d are coefficients that have fixed values for different classes of projects.
COCOMO model coefficients:
Software Project
Organic
2.4
1.05
2.5
0.38
Semi-detached
3.0
1.12
2.5
0.35
Embedded
3.6
1.20
2.5
0.32
Cyclomatic complexity V(G) for a flow graph G with V vertices and E edges is defined as,
V(G) = E V + 2
If software project cost is K and peak development time = td then,
Peak manning time,
If team size =n, then no. of communication channels =
Node X dominates node Y iff all the paths from initial node to node Y passes through node X.
Cyclomatic Testing comes under white box testing.
Span of control = number of immediate sub-routines.
Boundary value analysis and equivalence class partitioning both are test case design techniques in
black box testing.
For a function of n variables,
Boundary value analysis yields (4n + 1) test cases.
P a g e | 26
If T=life time of a software and initial development cost = x, then
Total cost of a project = Development cost + T (Development cost
Let us assume a software tested by error seeding technique.
Let N be the total number of defects in the system.
N be the number of these defects found by testing.
S be the total number of seeded defects.
S be the number of seeded defects found during testing
Then, the number of defects still remaining after testing =