You are on page 1of 53

Dr. D. Y .

Patil Institute of Technology, Pimpri, Pune -411018

Progress Seminar VI

PhD Topic: QoE Centric Network Engineering through Prediction


Framework

Name of the Candidate: Nilakshee Rajule

Name of the Research Guide: Dr. Mithra Venkatesan


Name of the Research Co-Guide: Dr. Radhika Menon

Research Center: Department of E&TC Engineering, DIT, Pimpri


Objectives
➢ To study the prediction or forecasting based on temporal dependency & spatial correlation among network
traffic to make accurate traffic forecast for a future period of time.

➢ To build energy efficient architecture for wireless networks (Cellular, WSN, IoT).

➢ To incorporate network traffic data in our architecture for prediction and analyzing effect on performance
parameters.

➢ To incorporate interference data in our architecture and analyzing effect on performance parameters.

➢ To incorporate user mobility and localization data in our architecture for prediction and analyzing effect on
performance parameters.

➢ To develop multi-objective network model for Intelligent Wireless Networks.

10/4/2023 Department of E&TC Engineering 2


Progress 1
Work Done Comments from experts

❑Carried out the study of network traffic ❑Satisfactory literature survey


forecasting methods for long-term traffic
❑Good presentation.
prediction.
❑Book chapter relevant to literature survey is
❑Identified the gaps in the literature survey.
published in CRC Press, Taylor & Francis
❑Identified the areas where predictive group.
analysis of network traffic can be utilized.

10/4/2023 Department of E&TC Engineering 3


Progress 2
Work Done Comments from experts

❑Developed a mathematical model for Base ❑Mathematical model for Base Station
Station switching strategy sleeping strategy has been developed.
❑Preprocessing of dataset for prediction ❑Satisfactory presentation.

10/4/2023 Department of E&TC Engineering 4


Progress 3
Work Done Comments from experts

❑Implementation of Prediction Module ❑Presentation is satisfactory


❑Linear Regression Technique
❑Publication in peer reviewed journal is
❑Artificial Neural Network expected.
❑Deep Learning
Model RMSE
ARIMA 603
Multi-Layer Perceptron 271.65
LSTM when data is fed
248.7
sequentially
LSTM when data is fed
71.6
parallelly

10/4/2023 Department of E&TC Engineering 5


Progress 4
Work Done Comments from experts
❑Implementation of traffic and interference- ❑Presentation is satisfactory.
aware BS Switching Technique ❑Suggested for implementation of various
❑Two types of traffic types i.e. Uniform and algorithms and comparison.
non-uniform traffic variations ❑Publication in peer reviewed journal is
expected
❑For each traffic type: implementation of 3
scenarios
Non-Uniform Traffic
Uniform Traffic Variation
Variation
Total Power Total Power Power
Power
Consumption Consumption Saving
Saving (%)
(Kwatt) (Kwatt) (%)

Scenario 1 1244.88 - 1806.6 -


Scenario 2 854.11 31.39 1272.14 29.58
Scenario3 843.9 32.21 1233.92 31.70

10/4/2023 Department of E&TC Engineering 6


Progress 5
Work Done Comments from experts
❑Traffic driven Centralized Cell zooming ❑Presentation by research scholar is good.
technique for power consumption reduction.
❑Has developed model as per problem
❑Reinforcement Learning based Mathematical statement.
model for cell zooming.
❑Published 2 papers and 1 book chapter in
❑Power Saving Reward
peer reviewed journal/conference.
❑Delay Reward/ Penalty
❑Interference Reduction Reward
❑Compared Power Saving of proposed technique
with existing work.
❑33.9% power saving is achieved with proposed
Cell zooming technique.
❑Proposed technique works effectively during low
load traffic conditions

10/4/2023 Department of E&TC Engineering 7


Work to be Presented in Progress 6
Distributed Cell Zooming Technique for Normal Traffic Load Conditions

01 02 03 04

Distributed Performance
Dynamic Load Zooming for Reward Parameters
Balancing Normal Load Calculations Evaluation
Conditions

Spider Monkey Reinforcement Rewards: Performance Parameters:


Algorithm for Learning for 1. Power Saving 1. Spectral Efficiency
Dynamic Load Cell Zooming 2. Delay 2. Energy Efficiency
Balancing Implementation 3. Interference Reduction 3. BS Utilization

10/4/2023 Department of E&TC Engineering 8


Introduction
➢ Rapid growth of the number of users in the cellular networks.

➢Traditional network infrastructure is unable to cope with this increased user traffic.

➢Ultra-dense HetNets are the useful solution.

➢Small cell Base Stations (SBS) can provide high data rates.

➢SBS increase the total power consumption in the cellular networks.

➢Need to develop an algorithm which will enhance the energy efficiency of the network without compromising
QoS.

10/4/2023 Department of E&TC Engineering 9


Energy Efficiency Approaches
Network Planning BS Density Optimization

& Deployment HetNet Deployment

Optimization of Beamforming Design


Radio Transmission • System-level approach
Resource Allocation
Process • Can be done with the same
network without hardware
Energy Efficiency BS Sleeping On-Off Switching Control replacement
Approaches Strategies Cell Zooming • Cells with low traffic are
switched to sleep mode.
RF Chain Design • Cell size is adjusted according to
Hardware Solutions dynamic traffic variation
Cloud-based Approach

Energy Harvesting Harvesting form Natural Sources

& Transfer Harvesting from RF

10/4/2023 Department of E&TC Engineering 10


Cell Zooming Techniques

Centralized Cell Zooming

Distributed Cell Zooming

10/4/2023 Department of E&TC Engineering 11


Literature Review
Paper Year Methodology/ Technique Contribution Limitations
Metaheuristic algorithms can be
Joint BS Sleeping and Combination of cell zooming and BS sleeping
[1] 2022 used for further enhancement of
Zooming Approach technique
power saving
Distributed Optimization Masked noise data approach for centralized and Real time traffic fluctuations are
[2] 2022 Approach distributed cell zooming not considered in the work.

Joint cell zooming and sleeping strategy Better understanding of traffic


[3] 2022 Game Theory Approach variations can help to enhance the
results

Traffic driven cell zooming 20% power consumption reduction is achieved Dynamic changes in the traffic
[4] 2021 technique should be identified.

Joint zooming and sleeping Combination of cell breathing and BS sleeping Solutions based on neural networks
[5] 2021 method technique can be adapted for power allocation

Feed Forward Back Classifier based cell zooming mechanism for Computation of BS capacity is
[6] 2019 Propagation Neural switching ON/OFF individual BS complex process
Network

10/4/2023 Department of E&TC Engineering 12


Gaps Identified and Solutions

Power saving
during normal Traffic load Dynamic load
Gaps Identified traffic is required fluctuation should balancing of user
to further enhance be traced exactly traffic is necessary
energy efficiency

Distributed zooming
Spider Monkey User mobility
technique is proposed
Optimization analysis is carried
Algorithm is used to out for traffic load
to reduce power Solutions
consumption during
balance the traffic analysis
normal load
load among BS

10/4/2023 Department of E&TC Engineering 13


Contributions

User Mobility Traffic Load Distributed Cell Generalized


Analysis Balancing Zooming Reward

User mobility Spider Monkey Reinforcement Balance between


analysis is carried Optimization Learning-based energy saving and
out for initial user Algorithm is used Distributed Cell preserving QoS in
association with BS for Load balancing zooming different traffic
conditions

10/4/2023 Department of E&TC Engineering 14


Self
Organizing
Energy
Efficient
Cellular
Network

10/4/2023 Department of E&TC Engineering 15


Proposed System

10/4/2023 Department of E&TC Engineering 16


Spider Monkey Optimization
• SMO mimics the foraging behavior of spider monkeys.
• These monkeys fall, in the category of fission-fusion social structure (FFSS) based animals.
• The FFSS of swarm may reduce the foraging competition among group members by dividing them into sub-groups
in order to search food.
• A female (global Leader) generally leads the group and is responsible for searching food sources.
• If she is not able to get enough food for the group, she divides the group into smaller subgroups that forage
independently.
• Sub-groups are also supposed to be leaded by a female (local leader) who becomes decision-maker for planning
an efficient foraging route each day.
• The group members communicate among themselves and with other group members, to maintain social bonds
and territorial boundaries.

10/4/2023 Department of E&TC Engineering 17


Why Swarm Intelligence?

Key Features of SMO


Division Self
of Labor
Organizat • Search based optimization
ion
• Dynamicity in the foraging behaviour.
• Offers a better balance between exploitation
and exploration
• High convergence Speed
• Inbuilt mechanism for stagnation check
Intelligent Decision Search-based

10/4/2023 Department of E&TC Engineering 18


SMO Phases Local Leader Phase

Global Leader Phase

Global Leader Learning Phase

Local Leader Learning Phase

Local Leader Decision Phase

Global Leader Decision Phase

10/4/2023 Department of E&TC Engineering 19


Network
Scenario

Macro Base Station(MBS): Global Leader

Small Base Station(SBS): Local Leader

10/4/2023 Department of E&TC Engineering 20


SMO for Load Balancing

Position update of users in Associate the users to the BS


Identify BS with low load
SBSs which are in active Mode

Global Local Local Global


Local Global
Leader Leader Leader Leader
Leader Leader
Learning Learning Decision Decision
Phase Phase
Phase Phase Phase Phase

Identify the high-speed and Select the non-adjacent Activate the BS in


low speed users. Analyze the SBSs and central SBS for hibernation according to
speed and direction of users Hibernation increased traffic load

10/4/2023 Department of E&TC Engineering 21


Local Leader Phase
Algorithm 1: Position Update Process in Local Leader Phase

for each user Ui ϵ kth BS do


for each k ϵ {1,2,…….M} do
if Q(0,1) ≥ pr then
𝑈𝑖+1 = 𝑈𝑖 + 𝑄 0,1 𝑋 𝑥𝑖+1 − 𝑥𝑖 + 𝑄 −1,1 𝑋(𝑦𝑖+1 − 𝑦𝑖 )
else
𝑈𝑖+1 = 𝑈𝑖
end if
end for
end for

10/4/2023 Department of E&TC Engineering 22


Global Leader Phase
Algorithm 2: User Mobility Analysis in Global Leader Phase (GLP)

for each user Ui ϵ kth BS do


for each k ϵ {1,2,…….M} do
if 𝑉𝑢𝑠𝑒𝑟 > 𝑉𝑡ℎ then
𝐴𝑠𝑠𝑜𝑐𝑖𝑎𝑡𝑒 𝑢𝑠𝑒𝑟 𝑤𝑖𝑡ℎ 𝑀𝐵𝑆
else
Analyse initial movement angle and velocity of user
end if
end for
end for

10/4/2023 Department of E&TC Engineering 23


Global Leader Learning Phase & Local Leader Learning Phase
Algorithm 3: Identification of SBS with low traffic load in Global Leader Learning Phase (GLLP)
Selecting SBSs for Hibernation in Local Leader Learning Phase (LLLP)

for each k ϵ {1,2,…….M} do

Calculate the available traffic load of SBSs


Sort the SBSs in ascending order of available traffic load
Identify the SBS with minimum load
end for
Select the SBS with low load, non adjacent SBSs and Central SBS to switch in
hibernation mode

10/4/2023 Department of E&TC Engineering 24


Local Leader Decision Phase & Global Leader Decision Phase

Local Lead Decision Phase:


User Association is done using load Balancing Algorithm (Slide:36)

Global Leader Decision Phase:


Activation of SBSs in hibernation mode according to increased traffic load using load
Balancing Algorithm (Slide:36)

10/4/2023 Department of E&TC Engineering 25


SMO Control Parameters

SMO Parameters Proposed System Parameters Value

LocalLeader Limit SBS limit 10

GlobalLeader Limit MBS limit 10

Maximum no. of BS required to


Maximum Groups (MG) N/120 (N is total no. of users)
Serve user traffic

Perturbation Rate Mobility Threshold Vuser ≠ 0

10/4/2023 Department of E&TC Engineering 26


Fitness Function
Fitness Calculation= Max (No. of BS with low traffic load)

Once a user association is determined, power consumption of the network is calculate as:

𝑃 = 𝑃𝑓𝑖𝑥𝑒𝑑 + 𝑃𝑑𝑦𝑛𝑎𝑚𝑖𝑐

The proposed network model is a heterogeneous network that consists of SBSs and MBS. Hence the total power
consumed by network can be expressed as:
𝑛
𝑠𝑤𝑖𝑡𝑐ℎ𝑖𝑛𝑔
𝑃𝑡𝑜𝑡𝑎𝑙 = 𝑃𝑚 + ෍ 𝑃𝑖𝑏 + 𝑃𝑖
𝑖=1

Where 𝑃𝑚 𝑖s power consumption of MBS,𝑃𝑏 𝑖s power consumption of SBS and 𝑃𝑠𝑤𝑖𝑡𝑐ℎ𝑖𝑛𝑔 𝑖s power consumed by BS
while switching from one state to the other.

Hence for the proposed work the objective function can be formulated as :

𝑃𝑡𝑜𝑡𝑎𝑙 < 𝑃𝑚𝑎𝑥

10/4/2023 Department of E&TC Engineering 27


Cell Zooming with Reinforcement Learning
BS States States & Actions
S1: BS in ON mode
S2: BS in Sleep Mode
S3: BS in Zoomed in Mode
S4: BS in Zoomed out Mode

Traffic Load &BS States


BS Load BS State
Zero Load State
Light Load State
Normal Load
Heavy Load

10/4/2023 Department of E&TC Engineering 28


Reward and Penalty

10/4/2023 Department of E&TC Engineering 29


Reward and Penalty
2

10/4/2023 Department of E&TC Engineering 30


Reward and Penalty

10/4/2023 Department of E&TC Engineering 31


BS Power Adjustment
Before Adjustment
Initial Power of BS: P1
Initial Radius of BS: R1
Signal Power received by user ‘a’ at
the coverage edge of BS is Pn1

After Power Adjustment


Transmitted Power of BS: P2
Adjusted Coverage Radius of BS: R2
Signal Power received by user ‘a’ at
the coverage edge of BS is Pm2

10/4/2023 Department of E&TC Engineering 32


Performance Parameters

𝑇ℎ𝑟𝑜𝑢𝑔ℎ𝑝𝑢𝑡 (𝑏𝑝𝑠)
𝑆𝑝𝑒𝑐𝑡𝑟𝑎𝑙 𝐸𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑐𝑦 =
𝐶ℎ𝑎𝑛𝑛𝑒𝑙 𝐵𝑎𝑛𝑑𝑤𝑖𝑑𝑡ℎ (𝐻𝑧)

𝑏𝑝𝑠
𝑆𝑝𝑒𝑐𝑡𝑟𝑎𝑙 𝐸𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑐𝑦 (
)
𝐸𝑛𝑒𝑟𝑔𝑦 𝐸𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑐𝑦 = 𝐻𝑧
𝑇𝑜𝑡𝑎𝑙 𝑝𝑜𝑤𝑒𝑟 𝐶𝑜𝑛𝑠𝑢𝑚𝑝𝑡𝑖𝑜𝑛(𝐽𝑜𝑢𝑙𝑒)

𝑁𝑜. 𝑜𝑓 𝐵𝑆 𝑖𝑛 𝑎𝑐𝑡𝑖𝑣𝑒 𝑚𝑜𝑑𝑒


𝐵𝑆 𝑈𝑡𝑖𝑙𝑖𝑧𝑎𝑡𝑖𝑜𝑛 =
𝑇𝑜𝑡𝑎𝑙 𝑛𝑜. 𝑜𝑓 𝐵𝑆𝑠

10/4/2023 Department of E&TC Engineering 33


User Mobility Analysis
• User k has an initial movement angle θ1,k that ranges
from 0 to 2π.

• θ2,k denotes the angle between user k and BS m.

• If angle between the movement angle of user k and the


connection between user k and BS m falls within the range
0 < |θ1,k − θ2,k| ≤ π/2, indicating that user k is moving
toward BS m.

• vk denotes the absolute velocity of user k.

• For each mobile user, the initial movement direction


remains unchanged for a period of time.

10/4/2023 Department of E&TC Engineering 34


Flowchart
for Load
Balancing

10/4/2023 Department of E&TC Engineering 35


Flowchart for Cell Zooming

10/4/2023 Department of E&TC Engineering 36


Simulation Parameters Type/Value
Network Heterogeneous
Parameters BS Type Capacity Cell-Macro Cell
Basic Cell- Small Cell
No. of Base Stations 7
No. of Users per BS Basic Cell-120 users
Macro Cell- 240 users
Total No. of Users 960
Power Consumption Small Cell: 25W
Macro Cell: 5.5KW
No. of Iterations 25
Signalling Period 5 Sec

10/4/2023 Department of E&TC Engineering 37


Power Saving Reward
Traffic % Average
Centralized Distributed
Probability Improvement %
0.38 0.686 0.743 5.692
0.45 0.450 0.699 24.885
0.53 0.315 0.519 20.412
0.71 0.241 0.424 18.253 14.008
0.74 0.187 0.325 13.731
0.75 0.157 0.257 10.007
0.8 0.136 0.187 5.074

Traffic Low Normal High

% Centralized 0.89 0.31 0.10


Improvement
Distributed - 0.45 -

10/4/2023 Department of E&TC Engineering 38


Delay Reward/Penalty
Traffic % Average
Centralized Distributed
Probability Improvement %
0.38 0.28 0.32 4.474
0.45 0.24 0.29 4.811
0.53 0.20 0.26 6.063
0.71 0.12 0.20 7.244 14.86
0.74 0.08 0.14 5.690
0.75 -0.23 0.02 25.757
0.8 -0.45 -0.32 12.230

Traffic Low Normal High

% Centralized 0.53 0.080 -0.69


Improvement
Distributed - 0.18 -

10/4/2023 Department of E&TC Engineering 39


Interference Reduction Reward
Traffic % Average
Centralized Distributed
Probability Improvement %
0.38 0.970 0.937 3.348
0.45 0.974 0.945 2.872
0.53 0.979 0.955 2.322
0.71 0.985 0.962 2.252 2.196
0.74 0.990 0.973 1.665
0.75 0.995 0.977 1.808
0.8 0.997 0.986 1.101

Traffic Low Normal High

% Centralized 4.61 1.56 0.23


Improvement
Distributed 3.76

10/4/2023 Department of E&TC Engineering 40


Base Station Utilization

• BS utilization of the centralized


zooming is more as all the SBSs works
in ON condition.
• Proposed SMO+RL Distributed
zooming aims to enhance the energy
efficiency by reducing the number of
active SBSs.
• BS utilization of proposed technique is
reduced by 3% compared to Joint
Sleeping and Zooming approach in [1].

10/4/2023 Department of E&TC Engineering 41


Energy Efficiency

• The Energy Efficiency of the proposed


SM+RL Distributed zooming technique
is more than the proposed centralized
zooming and Joint Sleeping and
Zooming approach in [1].
• The BS utilization is reduced in
proposed Distributed zooming.
• Energy efficiency is increased by 4.2 %
as compared to Joint Sleeping and
Zooming approach in [1].

10/4/2023 Department of E&TC Engineering 42


Spectral Efficiency

• Spectral Efficiency of centralized


zooming is more than other 2
techniques as all the SBSs are in ON
state.
• Transmit power of proposed SM +RL
technique is adjusted maintaining the
spectral efficiency in [1].
• Spectral Efficiency of proposed
SM+RL Distributed zooming is
increased by 1 % when compared with
Joint cell sleeping and zooming
approach in [1]

10/4/2023 Department of E&TC Engineering 43


Generalized Reward for Different Values of ‘α’

α=0.1 α=0.2 α=0.3 α=0.4 α=0.5 α=0.6 α=0.7 α=0.8


0.1 0.82 0.84 0.86 0.88 0.90 0.92 0.94 0.96
• If the value of α is close to 1, maintaining QoS
0.12 0.78 0.80 0.82 0.85 0.87 0.89 0.91 0.94 is prioritised over energy saving.
0.23 0.60 0.63 0.66 0.70 0.73 0.77 0.80 0.84
• As the value of α gets closer to 0, the agent
Average Traffic Probability

0.36 0.43 0.47 0.52 0.56 0.60 0.65 0.69 0.74


0.38 0.40 0.43 0.47 0.51 0.55 0.59 0.63 0.67
begins to embrace the proposed cell zooming
0.45 0.34 0.36 0.38 0.40 0.42 0.44 0.46 0.48
technique whenever it is feasible to save more
0.53 0.29 0.30 0.31 0.32 0.33 0.34 0.36 0.37
energy
0.71 0.22 0.23 0.25 0.26 0.27 0.28 0.29 0.30 • Dynamically setting α can balance between
0.74 0.18 0.19 0.20 0.21 0.22 0.23 0.25 0.26 energy saving and preserving QoS in different
0.75 -0.07 -0.03 0.01 0.04 0.08 0.12 0.16 0.20 traffic conditions.
0.8 -0.24 -0.19 -0.13 -0.07 -0.01 0.05 0.10 0.16
• α closer to 1 avoids high delaying and is
0.86 -0.39 -0.31 -0.24 -0.16 -0.08 -0.01 0.07 0.14
suitable for peak usage hours, while getting
0.91 -0.46 -0.38 -0.29 -0.21 -0.12 -0.04 0.05 0.13
closer to 0 saves more energy.
1 -0.69 -0.58 -0.46 -0.35 -0.24 -0.13 -0.01 0.10

10/4/2023 Department of E&TC Engineering 44


Comparison of Power Saving with Existing Literature
Average Power
Paper Technique
Saving
[1] Joint BS Sleeping and Zooming 25%

• Traffic driven cell zooming algorithm


[4] • Holt-Winters model for traffic prediction 20%

[7] • Metaheuristic Genetic Algorithm 28%

[11] • Self Organizing Energy Saving Technique 28.2%

Proposed Centralized • LSTM for traffic prediction


Zooming • Reinforcement Learning 33.91

Proposed Centralized + • SMO for load balancing


Distributed Zooming • Reinforcement Learning for cell zooming
37.41

10/4/2023 Department of E&TC Engineering 45


Conclusion
• A Distributed Zooming approach is presented for normal traffic load conditions.
• User mobility is analysed for initial association of user with SBS.
• For effective implementation of Distributed Zooming a SMO is used for balancing the user traffic load.
• Reinforcement Learning is used for implementing Distributed Cell zooming.
• Results show that 37.41% power can be saved by network when compared with always ON condition.
• When compared with existing literature [1] the proposed system offers:
• 4.3% increment in Energy Efficiency.
• 1% increment in Spectral Efficiency.
• 3% reduction BS utilization.

10/4/2023 Department of E&TC Engineering 46


Complete Work Flow
Objective Work Carried Out Contribution/ Results Outcome
Innovation

Identified the gaps


study of network traffic forecasting Book Chapter
in the literature
1: To study existing literature. methods for long-term traffic Published paper in
survey.
prediction. Taylor and Francis

2: To build energy efficient RMSE: 274 Published paper in


Developed mathematical model for Implementation of
architecture for wireless IJETT Scopus
energy efficient network. prediction model
networks indexed Journal

3: To incorporate network
RMSE Result:
traffic data in our architecture Implementation of traffic prediction Compared all 3
ARIMA Model: 603
for prediction and analyzing algorithm using ARIMA, ANN and models based on
ANN Model: 271.65
effect on performance LSTM RMSE
LSTM Model: 70.1 Paper accepted in
parameters.
IJCNIS Scopus
Two types of traffic Power Saving: indexed Journal
4: To incorporate interference
Implementation of traffic and types i.e. Uniform Uniform Traffic
data in our architecture and
interference-aware BS Sleeping and non-uniform distribution: 32.21%
analyzing effect on
technique traffic variations are Non-Uniform Traffic
performance parameters
considered distribution: 31.70%
10/4/2023 Department of E&TC Engineering 47
Complete Work Flow
Objective Work Carried Out Contribution/ Innovation Results Outcome

5: To incorporate user • Published survey


Implementation of • Network can save power upto
mobility and localization Reinforcement Learning paper in Scopus
Reinforcement 33.9% as compared to always
data in our architecture for based algorithm is used for indexed
Learning based ON condition.
prediction and analyzing implementation of cell Conference
algorithm for • The interference is reduced by
effect on performance zooming technique. • Paper Submitted
Centralized Cell 3% when compared with the
parameters to SCI Journal
Zooming. always ON condition

• Network can save power upto


• SMO-based load 37.41% as compared to always
6: To develop multi- Implementation of balancing for efficient ON condition.
objective network model SMO based load implementation of • When compared with Joint cell Paper will be
for Intelligent Wireless balancing and RL Distributed Cell Zooming sleeping and zooming approach submitted to SCI
Networks based distributed Cell • Generalized reward for in [1] Journal
Zooming Algorithm user preference ✓ EE increased by 4.3%
✓ SE increased by 1%
✓ BS utilization is reduced by 3%

10/4/2023 Department of E&TC Engineering 48


Publications
❑Research Papers: 03 published, 01 Accepted, 01 Submitted
1. Book chapter titled “Predictive Analysis of Cellular Networks: A Survey” published in Metaheuristic
Algorithms in Industry 4.0, CRC Press, Taylor & Francis Group.
2. Paper titled “Predictive Analysis Approach for Small Cell Base Station Sleeping Strategies” published in
International Journal of Engineering Trends and Technology (Indexing: Scopus).
3. Paper titled “Mobility Prediction in Cellular Networks: A Survey” published in International Conference of
Recent Trends Electronics & Communication (ICRTEC-2023) (Indexing: Scopus)
4. Paper titled “Network Traffic Prediction with Reduced Power Consumption towards Green Cellular
Networks” accepted in International Journal of Computer Network and Information Security(IJCNIS)
(Indexing: Scopus)
5. Paper titled “Reinforcement Learning based Self-Organizing Heterogeneous Ultra Dense Green Cellular
Networks” is submitted to IETE Journal of Research.

10/4/2023 Department of E&TC Engineering 49


Achievements
❑Won second prize in PhD category in online Symposium of Research Scholars (SYRES) jointly organized by
IEEE joint Chapter of AP, MTT, EMC (JCAME) & Communication Society, Pune section.

10/4/2023 Department of E&TC Engineering 50


References
[1] X. Xu, C. Yuan, W. Chen, X. Tao and Y. Sun, "Adaptive Cell Zooming and Sleeping for Green Heterogeneous Ultradense
Networks," in IEEE Transactions on Vehicular Technology, vol. 67, no. 2, pp. 1612-1621, Feb. 2018, doi:
10.1109/TVT.2017.2749058.
[2] M. Wakaiki, K. Suto and I. Masubuchi, "Cell Zooming With Masked Data for Off-Grid Small Cell Networks: Distributed
Optimization Approach," in IEEE Transactions on Control Systems Technology, vol. 30, no. 4, pp. 1696-1710, July 2022, doi:
10.1109/TCST.2021.3122164.
[3] Zhixin Liu, Xi Chen, Yi Yang, Kit Yan Chan, Yazhou Yuan, “Joint cell zooming and sleeping strategy in ultra dense
heterogeneous networks”, Computer Networks, Volume 220, 2023, 109482, ISSN 1389-1286,
https://doi.org/10.1016/j.comnet.2022.109482.
[4] Madhu Sudan Dahal, “Energy saving in 5G mobile communication through traffic driven cell zooming strategy”, Energy
Nexus,Volume 5,2022, 100040, ISSN 2772-4271, https://doi.org/10.1016/j.nexus.2022.100040.
[5] S. Habibi, V. Solouk and H. Kalbkhani, "Adaptive Sleeping Technique to Improve Energy Efficiency in Ultra-Dense
Heterogeneous Networks," 2019 5th Conference on Knowledge Based Engineering and Innovation (KBEI), Tehran, Iran, 2019,
pp. 782-786, doi: 10.1109/KBEI.2019.8734908.
[6] Siva Chandran, “ A Mathematical Model for Cell Zooming Mechanism of Base Station using Classification Approach”,
Applied Mathematics & Information Sciences 13(6):1053-1058, January 2019.

10/4/2023 Department of E&TC Engineering 51


References
[7] Md. Shahin Alom Shuvo, Md. Azad Rahaman Munna, Sujan Sarker, Tamal Adhikary, Md. Abdur Razzaque, Mohammad
Mehedi Hassan, Gianluca Aloi, Giancarlo Fortino, “Energy-efficient scheduling of small cells in 5G: A meta-heuristic
approach”, Journal of Network and Computer Applications, Volume 178, 2021, 102986, ISSN 1084-8045,
https://doi.org/10.1016/j.jnca.2021.102986.
[8] X. Yu, G. Li and W. Lu, "Power consumption based on 5G communication," 2021 IEEE 5th Information
Technology,Networking,Electronic and Automation Control Conference (ITNEC), Xi'an, China, 2021, pp. 910-914, doi:
10.1109/ITNEC52019.2021.9587128.
[9] Xu, X., Yuan, C., Chen, W., Tao, X., & Sun, Y. (2017). Adaptive cell zooming and sleeping for green heterogeneous
ultradense networks. IEEE Transactions on Vehicular Technology, 67(2), 1612–1621.
[10] G. A. Rizzo and M. A. Marsan, "The energy saving potential of static and adaptive resource provisioning in dense cellular
networks," 2018 10th International Conference on Communication Systems & Networks (COMSNETS), Bengaluru, India,
2018, pp. 297-304, doi: 10.1109/COMSNETS.2018.8328211.
[11] Z. Li, P. Yu, W. Li and X. Qiu, "Modeling and optimization of self-organizing energy-saving mechanism for HetNets,"
NOMS 2016 - 2016 IEEE/IFIP Network Operations and Management Symposium, Istanbul, Turkey, 2016, pp. 146-152, doi:
10.1109/NOMS.2016.7502807.

10/4/2023 Department of E&TC Engineering 52


Thank You

10/4/2023 Department of E&TC Engineering 53

You might also like