=Paper=
{{Paper
|id=Vol-2590/paper2
|storemode=property
|title=E2CT: Energy Efficient Cipher Technique
|pdfUrl=https://ceur-ws.org/Vol-2590/paper2.pdf
|volume=Vol-2590
|authors=Rahul Johari,Harshit Bhatia,Kalpana Gupta
|dblpUrl=https://dblp.org/rec/conf/micsecs/JohariBG19
}}
==E2CT: Energy Efficient Cipher Technique==
https://ceur-ws.org/Vol-2590/paper2.pdf
E2CT: Energy Efficient Cipher Technique
Harshit Bhatia 1, Rahul Johari 2, Kalpana Gupta 3
1
REVAL India Private Limited, Gurugram, India
2
SWINGER (Security, Wireless IoT Network Group of Engineering and Research) Lab,
USICT, GGSIP University, Sector-16C, Dwarka, Delhi, India
3
C-DAC, NOIDA, India
droid.harshit@gmail.com
rahuljohari@hotmail.com
kalpana7gupta@gmail.com
Abstract. The conventional techniques for symmetric and asymmetric cryptog-
raphy are not optimized for usage on handheld devices in their raw form. They
do not focus on optimized usage of battery over mobile devices and hence drain
significant battery when deployed over the wireless handheld devices. Fur-
thermore, they make use of a limited domain of keys and a limited number of
mathematical operations. The major portion of the existing traditional symmet-
ric cipher techniques is covered by those that rely on a single key-function for
generation of keys that are used to garble the plain-text to unintelligible text be-
fore sending it over an unsecure network. The increase in the number of encod-
ing operations and keys add significantly to the strength of a cryptographic
technique. This paper presents a power optimized symmetric key technique that
aims to reduce battery footprint without compromising on security by using
multiple keys coupled with multiple encryption operations.
Keywords: Green, Symmetric, Cryptography, Encryption, Decryption, Energy
Efficient.
1 Introduction
There exist plenty of cryptographic techniques that provide the security of the sensi-
tive data. [1, 2]. However, such traditional techniques were not aimed at catering to
the handheld devices with limited battery and resources. The techniques are not power
optimized and consume enormous amount of battery thus making them an unsuitable
choice for deployment in the mobile devices over the wireless network. The demand
for new and improved cryptographic techniques is high, especially for the intricate
hand-held devices that transmit sensitive information over network. The newer tech-
niques aimed for handheld devices need to be cheaper (in terms of battery consump-
tion) and faster without making any compromises with the security to ensure data
transmission at an overall lower energy cost. The technique proposed in this text is a
power optimized cryptographic approach aimed for mobile hand-held devices to se-
cure the data with minimal consumption of the energy and hardware resources. 1
Copyright © 2019 for this paper by its authors. Use permitted under Creative Commons Li-
cense Attribution 4.0 International (CC BY 4.0).
2
2 Proposed System
2.1 The cryptosystem
The proposed system introduces a lightweight cipher technique that aims at reduction
of overall battery consumption over a mobile handled device without making heavy
compromises on data security. The power optimized version of the predecessor, Pen-
taPlicative Cipher Technique, also uses a set of five predefined keys as an input to a
symmetric key cipher technique. However, unlike the PentaPlicative technique, the
power optimized version makes use of much cheaper mathematical operations –
XOR. In order to disguise the true length of plaintext from the sniffer, the power op-
timized technique also makes use of the bit-dispersion technique. The technique has
the strong grounds because of the multiple number of keys with multiple mathemati-
cal operations, which makes it difficult to decipher the plain text and hence effective-
ly decreasing the overall probability of the cipher text to get decrypted by anyone
other than the intended recipient.
2.2 Related Work
In [3] authors(s) present a new technique, the Cross-Language Cipher (CLCT) Tech-
nique, which is aimed at securing the plaintext data by character mapping. In [4] au-
thor(s) presented a new tool built in Java that demonstrates how the Dictionary attack
and Brute Force attacks are used to break the authentication and highlights the Injec-
tion via SQL. In [5] authors presented a rudimentary cipher technique that is aimed at
providing the security by employing three set of pre-defined keys in the process of
encryption and decryption. In [6] author(s) presented the Pentaplicative Cipher Tech-
nique which makes use of five keys to encrypt the plaintext input by the user. In [7]
author(s) had designed, implemented and evaluated a new Algorithm for Scheduling
that makes use of a neural network predictor model to power off the unused servers in
a Cloud computing environment and essentially reducing the consumption of power.
In [8] author(s) propose various queuing algorithms to utilize the resources and task
assignment is done in an efficient manner to ensure that the overall cost of operations
is reduced while also making it a cleaner approach by decreasing the ill-effects of the
data center on environment resulting in Green Eco System.
3 Methodology
The Energy Efficient extension of PentaPlicative cipher technique aims at being
readily used in mobile handheld devices. This expects the technique to be stricter
when it comes to resource consumption and in turn battery consumption. Considering
this, Energy Efficient Cipher technique makes use of caching principle to store the
keys to cut down the dependency of gathering keys from device’s GPS and physical
address for every operation performed. Furthermore, the technique is lightweight
without making significant compromises with the security.
3
The GPS along with the mobile network data consumes a considerable amount of
battery from device, hence limiting the pings made to them for grabbing the location
helps in reduction of battery usage by a noticeable amount. Biggest culprits in faster
battery discharge is Screen brightness and CPU usage [9]. The PentaPlicative Cipher
technique caters to this problem of on-screen time by reduction of screen brightness
during the process. The CPU and memory usage have been notably reduced by em-
ploying garbage collection to free up memory and killing any lingering daemon
threads upon completion of encryption/decryption process, thereby making an ob-
servable reduction in the overall battery consumption by the handheld device.
3.1 Key Arrangement
The Energy Efficient Cipher technique makes use of a set of five private keys (kept
secret) which are derived from the physical information of the handheld device. The
set of keys is unique to every device; hence authentication of sender can easily be
made. The IMEI number along with the location coordinates form the input key vec-
tor. The five keys (K1, K2, K3, K4 and K5) are derived from the input domain as the
following text describes. The unique International Mobile Equipment Identity (IMEI)
number forms the first three keys (K1, K2, K3) with each key of length five integer
numbers and the other two keys (K4, K5) are given by first five integers (without dec-
imal) derived from latitude and longitude coordinates respectively. The length of keys
has been fixed to five to make computations faster and cheaper in terms of resource
consumption. This allows for an optimized lightweight technique that would limit the
actual battery discharge.
Additionally, using the physical information available on devices decreases the
human intervention to bare minimum in the complete process, thus enabling the tech-
nique to be easily plugged in with other data storage or transmission applications that
require data encryption.
The caching mechanism is also embedded in the key generation process. This is
necessary because the key generation relies heavily on the physical information of the
device and this would require the technique to make connection to the network for
grabbing the location coordinates and reading of the hardware chipset for the IMEI
number. This entire process wastes many precious CPU cycles as well as adds an
additional burden by using the heavy battery sucking resources of the device. The
technique makes sure to close the network connections if there are cached keys avail-
able to it and hence saving an appreciable number of unnecessary pings that would
have been wasted otherwise.
3.2 Encryption operation
The paper introduces the technique with an example that draws by taking a sample
plaintext and sample keys portraying the encryption and decryption on input domain.
The technique, like its predecessor, also makes use of the ASCII character set. The
input characters of the plaintext are first converted into the decimal numbers by mak-
4
ing use of the ASCII character to decimal conversion and then further mathematical
operations are performed on them with the set of five input keys.
In order to mask the true length of the Plaintext (PT), the technique makes use of an
operation called the “Bit-Dispersion”. This operation first converts the ASCII charac-
ters to their corresponding decimal values and then converts the decimal values to a
Base2 Binary number. Each decimal number is thus converted to an 8-digit binary
number. These 8-digit binary numbers are grouped together to form a long stream of
binaries. The function further makes a group of 6 bits from this stream and then con-
verts this -bit binary number to the corresponding ASCII character. If there are any
remainder bits which are left after the grouping, are appending with padding of zeroes
to make it a 6-bit binary number and this is converted to the corresponding ASCII
character too. This new set of ASCII characters will be the final Cipher Text (CT)
which would be of a different length as the original Plaintext.
There are five other pre-defined mathematical operations other than the Bit-
Dispersion that the technique uses. These mathematical operations when performed in
a sequential manner would result in the Ciphertext (CT) which can then be sent out by
the sender to the receiver. These operations, denoted by E1, E2, E3, E4, E5, are as fol-
lows:
E1 = (PT XOR K1) (1)
E2 = (E1 + K2) mod 256 (2)
E3 = (E2 * K3) mod 256 (3)
E4 = (E3 - K4) mod 256 (4)
E5 = (E4 XOR K5) (5)
CT = bit dispersion (E5) (6)
The table 1and 2 depicts the process of encryption with the help of an example. The
example clearly defines the input domain of plaintext and secret set of keys, followed
by the set of encoding operations. Please note that for this example the keys that are
selected are very small and simplistic numbers to ease the demonstration of mathe-
matical operations, but in the practical world much larger keys would be used.
Plaintext (PT) - CIPHER
Let the private keys be:
K1 = 17
K2 = 19
K3 = 17
K4 = 13
K5 = 15
5
Table 1. Encryption Table
PT E1= (P.T. E2= (E1 + K2) E3= (E2 * K3) E4= (E3 – K4) E5 = (E4
XOR K1) mod 256 mod 256 mod 256 XOR K5)
C(67) (67 XOR 17) (82 + 29) mod (111 * 13) mod (163 - 57) mod (106 XOR 19)
= 82 (R) 256 = 111 (o) 256 = 163 (ú) 26 = 106 (j) = 121 (y)
I(73) (73 XOR 17) (88 + 29) mod (117 * 13) mod (241 - 57) mod (184 XOR 19)
= 88 (X) 256 = 117 (u) 256 = 241 (±) 256 = 184 (©) = 171 (½)
P(80) (80 XOR 17) (65 + 29) mod (94 * 13) mod (198 - 57) mod (141 XOR 19)
= 65 (A) 256 = 94 (^) 256 = 198 (ã) 256 = 141 (ì) = 158 (×)
H(72) (72 XOR 17) (89 + 29) mod (118 * 13) mod (254 - 57) mod (197 XOR 19)
= 89 (Y) 256 = 118 (v) 256 = 254 (■) 256 = 197 (┼) = 214 (Í)
E(69) (69 XOR 17) (84 + 29) mod (113 * 13) mod (189 - 57) mod (132 XOR 19)
= 84 (T) 256 = 113 (q) 256 = 189 (¢) 256 = 132 (ä) = 151 (ù)
R(82) (82 XOR 17) (67 + 29) mod (96 * 13) mod (224 - 57) mod (167 XOR 19)
= 67 (C) 256 = 96 (`) 256 = 224 (Ó) 256 = 167 (º) = 180 (┤)
Table 2. Bit dispersion Operation
Ob-
121 171 158 214 151 180
tained E5
E5 bi- 01111 10101 10011 11010 10010 10110
nary 001 011 110 110 111 100
Ci- 01111 01101 10111 01111 11010 10100 01 11
pher 0 0 0 0 1 1 0001 0100
Ci-
36 32 56 36 65 51 36 64
pher Text
Final transmitted Cipher text for ‘CIPHER’ plaintext is $ 8$A3$@
3.3 Decryption operation
The Ciphertext received at the receiver’s end needs to be converted to the actual plain
text message and this process is called the decryption. In order to decrypt the garbled
cipher text message, the receiver also uses the same set of keys that sender used to
encrypt the message. The first step is to change the length of the cipher text to match
the original length of the plain text message. The Bit-dispersion operation that sender
performed needs to be neutralized by the reverse bit-dispersion mechanism. This pro-
cess now re-groups the 6-bit binary characters to the 8-bit binary characters and then
converts the 8-bit binary to the corresponding ASCII Character. The extra padding
bits in the form of zeroes, that were added during the Bit-Dispersion process are also
6
removed and the original length of plaintext is restored on received cipher text. This is
followed by the set of pre-defined mathematical operations using the set of private
keys to count the effects of encryption to finally obtain the original desired plaintext
message. The mathematical steps are denoted as D1, D2, D3, D4 and Dc denoted the
reverse bit-dispersion operation. The mathematical operations to compute the
plaintext (PT) on the receiver’s end are depicted as follows:
Dc = reverse bit dispersion (CT) (7)
D1 = (Dc XOR K5) (8)
D2 = (D1 + K4) mod 256 (9)
D3 = (D1 * K3-1) mod 256 (10)
D4 = (D3 – K2) mod 256 (11)
PT = (D4 XOR K1) (12)
The tables 3 and 4 depict the usage of the mathematical operations in the decryption
process by making use of an example. The example depicted here is an extension of
the same example depicted in Tables 1 and 2. The decryption operation takes in the
input the same set of five keys as private keys and also uses the output of encryption
operation as the input of decryption operation as a cipher text.
The Cipher text (CT) is: $ 8$A3$@
The modulo inverse of the Key K3 is denoted as K3-1 and is computed to be: 197 (sat-
isfies K3 K3-1 ≡ 1 mod 256)
Table 3. Reverse Bit Dispersion Operation
Ob-
36 32 56 36 65 51 36 64
tained C
C in 01111 01101 10111 01111 11010 10100 01 11
binary 0 0 0 0 1 1 0001 0100
Re- 01111 10101 10011 11010 10010 10110
Dispersed 001 011 110 110 111 100
Dis-
persed 121 171 158 214 151 180
ASCII
Dispersed text to be used to obtain plaintext is y½×Íù┤
7
Table 4. Decryption Table
Dc D1 = (C XOR D2 = (D1 + K4) D3 = (D2 * K3-1) D4= (D3 - K2) D5= (D4
K5 ) mod 256 mod 256 mod 256 XOR K1)
y (121 XOR 19) (106 + 57) mod (163 * 197) mod (111 - 29) mod (82 XOR 17)
(121) = 106 (j) 256 = 163 (ú) 256 = 111 (o) 256 = 82 (R) = 67 (C)
½ (171 XOR 19) (184 + 57) mod (241 * 197) mod (117 - 29) mod (88 XOR 17)
(171) = 184 (©) 256 = 241 (±) 256 = 117 (u) 256 = 88 (X) = 73 (I)
× (158 XOR 19) (241 + 57) mod (198 * 197) mod (94 - 29) mod (65 XOR 17)
(158) = 241 (ì) 256 = 198 (ã) 256 = 94 (^) 256 = 65 (A) = 80 (P)
Í (214 XOR 19) (197 + 57) mod (254 * 197) mod (118 - 29) mod (89XOR 17)
(214) = 197 (┼) 256 = 254 (■) 256 = 118 (v) 256 = 89 (Y) = 72 (H)
ù (151 XOR 19) (132 + 57) mod (189 * 197) mod (113 - 29) mod (84 XOR 17)
(151) = 132 (ä) 256 = 189 (¢) 256 = 113 (q) 256 = 84 (T) = 69 (E)
┤ (180 XOR 19) (167 + 57) mod (224 * 197) mod (197 - 29) mod (67 XOR 17)
(180) = 167 (º) 256 = 224 (Ó) 256 = 96 (`) 256 = 67 (C) = 82 (R)
Final intended plain text message is ‘CIPHER’
4 Mathematical Modelling
The mathematical operations cannot be applied directly to the plain text which is a
string of characters. Before the encryption operations may be applied, the plain text
character needs to be converted from the text format to the corresponding ASCII dec-
imal number value. Upon this ASCII decimal encoding, the encryption operation can
be summarized as a set of mathematical equations which when applied in the correct
order result in the final cipher text which can then be transmitted to the receiver. Each
individual encryption equation from the set of mathematical equations can be denoted
by En(x); and each equation when applied on the plain text, denoted by P(x), outputs
the final Cipher text which is denoted by C(x) as follows:
C(x) = fdispersion (E5(x)) (13)
where, E5(x) = (E4(x) XOR K5(x)), (14)
and, E4(x) = (E3(x) - K4(x)) mod 256, (15)
and, E3(x) = (E2(x) * K3(x)) mod 256, (16)
and, E2(x) = (E1(x) + K2(x)) mod 256, (17)
and, E2(x) = (P(x) XOR K1(x)) (18)
The Bit-dispersion function in the above equations, is given by the function fdis-
persion(En(x)) and the private keys are given by the function Kn(x), where n 𝜖 [1, 5].
The length of the plain text is given as ‘n’ and that of cipher text is denoted by ‘m’
8
where n < m, since the bit dispersion function eliminates the one-to-one character
mapping and thus changes the length of final cipher text.
1. The conversion of the string of characters into their corresponding ASCII decimal
numbers is the first operation which is performed. This encoding of plan text char-
acters can be represented as a function P(x), where P(x) comprises of individual
decimal values and each of these decimal value for ‘n’ number of characters of
plain text can be represented as P1(x) P2(x) P3(x) … Pn(x) and each character Pi(x)
represented in Base10 decimal value belongs to the range 0 ≤ Pi(x) ≤ 255.
2. Each encryption operation is a linear mathematical operation which involves the
five private keys and each encryption mathematical operation can be represented as
the function E(x) comprised of a mathematical operation (represented as 𝜙) and a
key function K(x) and is represented as, E i(x) = Ei-1(x) 𝜙 Ki(x)
3. The Base10 decimal number that is obtained from the encryption functions E 4(x)
needs to be converted to a Base2 binary number. This Decimal to Binary conver-
sion is carried out for each individual decimal number from the set of ‘n’ numbers
and for a decimal number represented as xi the binary number can be obtained as a
set of following procedure – “Keep dividing the quotient by 2 until the quotient is
0 and the all the remainder represented in a reverse order is the binary number”.
This can be illustrated as a set of equations:
Q0 = xi / 2 remainder R0 (19)
Q1 = Q0 / 2 remainder R1 (20)
Qj = Qj-1 /2 remainder Rj, Qj 𝜖 [1, 0] (21)
Qj+1 = Qj / 2 remainder Rj+1, Rj+1 𝜖 [1, 0] (22)
The Base2 binary number representation of decimal integer x i is Rj+1Rj …. R2
R1
4. The bit-dispersion function fdispersion groups all the 8-bit binary numbers together
and then combine them into a 6-bit binary number which is then converted back to
the Base10 decimal number. The binary numbers need to be converted back to the
decimal numbers and this Base2 to Base10 conversion involves, “multiplying the
sum total by 2 and adding the remainder bit to it” and is shown as follows for a bi-
nary number Rj+1Rj …. R2R1 the final Decimal number is Dj+1:
D1 = 2 x 0 + R1 (23)
D2 = 2 x D1 + R2 (24)
Dj = 2 x Dj-1 + Rj (25)
Dj+1 = 2 x Dj + Rj+1 (26)
5. Conclusively, these transformed decimal integers obtained as the result of encryp-
tion operation E5(x) is mapped to an ASCII character each and this reverse ASCII
9
mapping gives the final Cipher text C(x) which is then returned to the receiver as
the message. This transmitted cipher text is of length ‘m’ which is greater than the
length of plaintext ‘n’, i.e. m > n.
6. The average execution time is given by equation, T = (∆T0 +∆T1 +∆T2+ ∆T3+ ∆T4+
∆T5+ ∆T6+ ∆T7+∆T8+ ∆T9+∆T10) / 11.
7. The Time complexity can be computed and depicted in Big-Oh notation as ‘O(n)’
where ‘n’ is the length of the plaintext. The calculation of the Time taken for vari-
ous process is specified in table 5.
Table 5. Time Calculation.
S. No. Operations Time taken
1. E1(y) ∆T0
2. E2(y) ∆T1
3. E3(y) ∆T2
4. E4(y) ∆T3
5. E5(y) ∆T4
6. C(y) ∆T5
7. ASCII convert ∆T6
8. Base10 to Base2 ∆T7
9. Bit Dispersion ∆T8
10. Base2 to Base10 ∆T9
11. Reverse ASCII ∆T10
10
Fig. 1. Battery Usage of PentaPlicative vs Affine cipher
Fig. 1. Discharge Speed - PentaPlicative vs Affine cipher
Fig. 2. Battery Usage Foreground and Background of PentaPlicative vs Affine Cipher
11
Fig. 3. CPU Usage of PentaPlicative vs Affine Cipher
Fig. 5. Flowchart of Encryption in PentaPlicative Cipher
5 Results: PentaPlicative vs Affine Cipher Technique
A comparison is drawn between two symmetric key techniques, PentaPlicative Cipher
technique and Affine Cipher technique. The PentaPlicative Cipher technique uses five
keys as opposed to the two-key cryptosystem of affine cipher. The use of limited keys
makes the affine cipher highly vulnerable to attacks that make use of a system of line-
ar equations to decipher the text. However, the use of increased number of keys and
operations coupled with the bit dispersion operation used in PentaPlicative cipher
technique makes it impenetrable to such attacks. It would be expected that since af-
fine cipher uses lesser keys and lesser operations, hence it would also result in a lower
consumption of battery than PentaPlicative. However, this would have been true if no
12
battery discharge optimizations were not included for PentaPlicative technique. It
makes use of cached keys thus reducing the hefty battery consumption as opposed to
the standard implementation of the affine cipher with the same input vector of keys of
IMEI number and location coordinates. The PentaPlicative also reduces screen
brightness and prompts garbage collection upon exit, thus making a significant im-
provement in effectively curbing memory as well as battery utilization. Battery usage
and discharge speed were measured for both PentaPlicative and Affine Cipher using
AccBattery App [10]. It reveals that battery usage for Affine Cipher is relatively
higher than PentaPlicative (Fig 1) and the former has a greater battery discharge speed
over the period (Fig 2). The CPU Usage along with foreground and background bat-
tery usage has also been included for the comparative analysis of the two apps pro-
vided by GSamBattery [11] and TrepnProfiler [12]. It indicates a greater foreground
task and higher usage of CPU in the background and as result a larger resource utili-
zation for Affine Cipher (Fig 3 and 4). The obtained results have been briefed in the
mentioned table. Please note that the simulation of the Energy Efficient Cipher Tech-
nique was carried out in a controlled environment (Table. 6). Both techniques were
coded as android applications and the performance results were measured under the
same conditions. The running time of Pentaplicative Cipher Technique is 1.9 milli-
seconds. [6]
Table 6. Simulation environment
Simulation Environment
O.S. used Android 8.0.0
Mobile Model OnePlus 3T (A3003)
RAM 6 GB
Development IDE Android Studio 2.3.1
Compile SDK Version 25
Development Lang Java, XML
Java Version 1.8.0 build 221-b11
6 Conlusion
The Energy Efficient Cipher technique is a Green Cryptographic technique that is
robust and lightweight to cater to the handheld devices. It has been optimized to make
minimal use of the hardware resources and to use light-weight operations to secure
the data. This, in turn, reduces the battery consumption by the Energy Efficient Ci-
pher Technique. The additional mechanisms implemented in the crypto system in
form of caching greatly reduce the number of CPU cycles and thereby significantly
reducing the battery consumption and making it a perfect fit for the hand-held devic-
13
es. Moreover, the reduction of human interaction in the process of the selection of
private keys by automatically selecting the five private keys as the physical infor-
mation of the device, contributes to the security and make it much more difficult for
the keys to be deciphered. Furthermore, the obtained results clearly support the Ener-
gy Efficient Cipher Technique in terms of the battery drainage and CPU usage over a
less secure traditional affine cipher technique. Conclusively, the result is a strong
green cipher technique.
References
1. Behrouz A. Forouzan, “Cryptography and Network Security”, Tata McGraw-Hill Special
Indian Edition 2007.
2. William Stallings. “Cryptography and Network Security-Principles and Practices” Pearson
Education, fourth Edition 2007.
3. L. Singh, R. Johari. "CLCT: Cross Language Cipher Technique." In International Sympo-
sium on Security in Computing and Communication, pp. 217-227. Springer International
Publishing, 2015.
4. I. Jain, R. Johari, R.L Ujjwal “CAVEAT: Credit Card Vulnerability Exhibition and Au-
thentication Tool”. In: Second International Symposium on Security in Computing and
Communications (SSCC‟14), pp 391- 399.Springer 2014.
5. R. Johari, H. Bhatia, S. Singh and M. Chauhan. “Triplicative Cipher Technique” Procedia
Computer Science 78: 217-223, 2016.
6. Garg N., Bhatia H., Johari R. (2019) Pentaplicative Cipher Technique. In: Bhattacharyya
S., Hassanien A., Gupta D., Khanna A., Pan I. (eds) International Conference on Innova-
tive Computing and Communications. Lecture Notes in Networks and Systems, vol 55.
Springer, Singapore.
7. Duy, Truong Vinh Truong, Yukinori Sato, and Yasushi Inoguchi. "Performance evaluation
of a green scheduling algorithm for energy savings in cloud computing." In Parallel &
Distributed Processing, Workshops and Phd Forum (IPDPSW), 2010 IEEE International
Symposium on, pp. 1-8. IEEE, 2010.
8. Sanjeevi, P., and P. Viswanathan. "A green energy optimized scheduling algorithm for
cloud data centers." In Computing and Network Communications (CoCoNet), 2015 Inter-
national Conference on, pp. 941-945. IEEE, 2015.
9. Shye, Alex, Benjamin Scholbrock, and Gokhan Memik. "Into the wild: studying real user
activity patterns to guide power optimizations for mobile architectures." In Microarchitec-
ture, 2009. MICRO-42. 42nd Annual IEEE/ACM International Symposium on, pp. 168-
178. IEEE, 2009.
10. AccuBattery Application downloaded from
https://play.google.com/store/apps/details?id=com.digibites.accubattery&hl=en
11. GSamBattery Application downloaded from
https://play.google.com/store/apps/details?id=com.gsamlabs.bbm&hl=en
12. TrepnProfiler Application downloaded from
https://play.google.com/store/apps/details?id=com.quicinc.trepn&hl=en