See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/381947332

An Efficient Food Distribution and Logistics Based on Supply Chain

Conference Paper · May 2024

DOI: 10.1109/AIIoT58432.2024.10574746

CITATION READS

1 116

4 authors, including:

Mounika Nalluri
Cigna
29 PUBLICATIONS 104 CITATIONS

SEE PROFILE

All content following this page was uploaded by Mounika Nalluri on 03 July 2024.

The user has requested enhancement of the downloaded file.

 2024 3rd IEEE International Conference on Artificial Intelligence for Internet of Things (AIIoT 2024)

An Efficient Food Distribution and Logistics Based

on Supply Chain
Chinna Babu Mupparaju Mounika Nalluri Aruna Sri Rongali
Department of Information Technology Department of Information Technology Department of Information Technology
2024 3rd International Conference on Artificial Intelligence For Internet of Things (AIIoT) | 979-8-3503-7212-0/24/$31.00 ©2024 IEEE | DOI: 10.1109/AIIoT58432.2024.10574746

University of Central Missouri Murray State University University of the Cumberlands

Missouri, USA Murray, USA Williamsburg, USA
[email protected] [email protected] [email protected]

Bhuvanesh Ananthan
Department of Electrical and Electronics Engineering
PSN College of Engineering and Technology
Tamil Nadu, India
[email protected]

particular benefit brings anguish to numerous organizations in

Abstract—Based on entire world, an important business the sector that cannot keep growing with these developments.
Production companies have faced significant difficulties as a
to different developed industries are handle of industry result of the swift shifts in customer appetite for their products
and food processing, which leads to higher on the world scale and the ambiguity around commercial
employability. A smooth food products packing and counterparts' financial objectives [3]. Businesses understand
production execution are the important roles in human that it is nearly hard to obtain an edge in the current economic
workforce. The lacking in food safety and supply chain climate on their own. Businesses need to strengthen
maintenance are filed in food industries because of communication and collaboration to mitigate hazards, jointly
create, and boost profitability.
human involvements. This work presented a novel
optimization as well as deep learning model for food The supply chain framework comprises interconnected
production and maintenance with supply chain components like administration, cash movement, and
management. Based on greenhouses, the production of knowledge movement throughout suppliers, whereas the
supply chain primarily concentrates on how to organize all the
foods like plant growth and yield is predicted using a participants to operate collectively as one entity to achieve
deep auto-encoder (DAE) model followed by the identical targets [4]. An organization's capacity to manage
retailing of food refrigeration model is explained. Next, material effectively, save expenses, boost capacity utilization,
the supply chain is optimized with Adaptive enhance customer and vendor contentment, and optimize
Waterwheel plant (AWP) algorithm to verify the expiry storage facilities are all indicators of a successful mobility
date and optical recognition in which the people health operation. It can cost firms a lot of money because this can
result in abandoned or rescheduled commands, fabrication,
and food safety are ensured. The proposed work decreased revenue, and customer share. Modifications can
performance is superior to other conventional also harm a business's image and may cause dissatisfaction
approaches. among suppliers and stakeholders. This work presented a new
Keywords—Food production, Supply chain, Retailing of food deep deep auto-encoder and adaptive waterwheel plant
refrigeration, Deep auto-encoder and Adaptive Waterwheel plant algorithm for ensuring an effective distribution of food and
algorithm. logistics with respect to optimized supply chain.
I. INTRODUCTION Rest of the article is arranged like; Section 2 pointed out
the existing works based on food supply chain management.
One of the primary corporate responsibilities that oversees
Section 3 describes the proposed methodology and its results
the flow of items and offerings amongst many different parties
are explained in Section 4. At last, the Section 5 concludes the
is supply chain management [1]. The term supply chain refers
whole work.
to the actions, entities, individuals, and assets associated with
converting initial supplies into final goods and finishing or II. RELATED WORKS
completing client orders. Numerous challenges, such as a
shortage of accountability, interruptions, additional Bhutta et al. [5] have presented the supply chain
complications, inaccurate data, and uncertainty, enhance the management (SCM) procedure to improve the supply network
difficulties of current supply networks. Logistics optimization collaborators' collaboration. These devices gather information
is achieved through managing market forces, and successful that is pertinent to the nutrition transportation framework, like
warehouse administration is a key component of this process recognizing individuals or items that have tags attached to
[2]. The adoption of innovative technologies and the rapid them and detecting environmental changes. Because it
expansion in knowledge accessibility create efficient stock misidentifies variable contexts, organizations run the risk of
and replacement procedures, which greatly enhance the damaging or even destroying the good. It intends to establish
transmission of data and supplies throughout the production a reliable tracking and communication platform that utilizes
chain. the Internet of Things. Therefore, it is outdated, slow, and
ineffective.
Integrating technology improves logistics performance by
increasing resource manufacturing and distribution. This

979-8-3503-7212-0/24/$31.00 ©2024 IEEE

Authorized licensed use limited to: University of the Cumberlands. Downloaded on July 03,2024 at 19:26:11 UTC from IEEE Xplore. Restrictions apply.
 Omar et al. [6] have described Vendor Managed Inventory including mapping precision, interruption, implementation
(VMI) within the procedures of the supply chain. Enhanced duration, and commuting duration, as part of a thorough
interaction and honesty among vendors can be achieved by examination. Thus, it is inappropriate to be employed in a
developing a blockchain-enabled VMI system to deliver setting where there is an opportunity for interpersonal
goods. It guarantees a solid economic alliance and raises interaction.
consumer trust. Additionally, it increases ordering precision
while preserving the confidentiality of information and may Zhiwen et al. [9] highlighted Supply chain logistics
result in higher revenues for all supply chain participants. information collaboration (SCLIC) is employed to examine
Hence, it is insufficient to satisfy the various demands of the the fluctuation in the total number of participants in source and
logistics field. manufacturing subgroups. To achieve SCLIC and raise the
sector's long-term marketability, a supply chain executive can
Peña et al. [7] have implemented food supply chain devise targeted initiatives to enhance the cooperative
management (FSCM) to facilitate safe exchanges across atmosphere for logistical knowledge and inspire distributors
various industries, and possess the attributes of distributed to adopt a cooperative manner of distributing the data.
computing, confidentiality of data, and dependability, in Enhancing overall transport knowledge, cooperation
response to such requirements. Choosing made by distributors perception, and productive conduct are beneficial. But it's not
on the amount of merchandise that buyers and sellers should always that simple, particularly in supply networks that lack
allocate can be made better with the use of collaborative data. the key components.
This sort of technology has the potential to enhance
communication and goods movement between every single From the detailed survey, the following challenges are
part if it is implemented as the foundation of a supply network. identified in food industry. Complexity of Supply Chains;
However, the techniques cannot keep up with the innovative Food Safety and Quality Control; Globalization and
standards. Geographic Diversity; Supply Chain Disruptions;
Sustainability and Environmental Concerns.
Nagarajan et al. [8] suggested an Internet of Things (IoT)
in urban areas, this infrastructure facilitates the immediate III. PROPOSED METHODOLOGY
agricultural sector's monitoring, analysis, and management. This section presents the overall schematic structure of
The nutritional value is guaranteed by the dynamically proposed model as depicted in Figure 1. Major fundamental
generated aliment logistics for urban areas, and it additionally factors of this block is productivity of food, maintenance with
offers sophisticated truck scheduling and the ability to identify storage and distribution of food. The detailed steps are
the locations of infection in the alimentary chain management. deliberated in the subsequent sub-section.
It is assessed using a wide range of efficiency measures,

Fig. 1. Proposed schematic structure

A. Prediction of plant growth environmental data. This work suggested a novel deep auto-
encoder (DAE) model to predict the growth of plant and well
Based on the research Centre of ornamental plant in as yield. An effective input data learning is auto-encoder that
Belgium, the Ficus plants growth and productivity is predicted reduces the dimension of data. When the hidden layer contains
via proposed deep auto-encoder (DAE) model. Through less nodes than the layers of input or output, the input
window opening, the adding of CO2, light assimilation, air representation compresses to treat the final hidden layer as
heating are controlled and monitored of greenhouse activation function. An auto-encoders applies to the activation
microclimate. Based on hours, continuously record the function. The input layer with the similar number of nodes

Authorized licensed use limited to: University of the Cumberlands. Downloaded on July 03,2024 at 19:26:11 UTC from IEEE Xplore. Restrictions apply.
contains auto-encoder of output layer. The input y is
` determining the input, complete the model training. The
reconstruction Z and the original Y based on the loss function
reconstructed to generate the supervised learning of y input
plant data provided. as l (Y , Z ) . Use various loss functions with the objective
squared error is computed.
Three iterative steps with back-propagation is the auto- 2
encoder training process. At the output layer, an output is l (Y , Z ) = Y − Z (1)
generated with the hidden layers after the activation of all
nodes data are calculated [10]. The link weights are updated In green house environment, there is complex and
and the error calculates the back-propagation. The classical dynamic structure of vegetables growth. In each environment,
auto-encoder develops the auto-encoder to construct the required calibration and initial parametric values along with
vigorous representation from higher dimensionality data. DAE to estimates the complexities. The DAE model is
Corrupt the input Y as input to perform p process. The process adopted to predict the vegetable productivities in UK in which
gθ map the corrupted input. The reconstruction of Y is the data collection includes yield actual measurement, inside
and outside temperature, radiation and humidity. DAE predict
generated to build the Y and conduct the procedure of g θ . The the growth of each plants based on these data as well as the
shuffled inputs plant resample the loss function is minimized productivities are measured on weekly basis. The Proposed
and target the auto-encoder process of training. While DAE to predict plant growth and productivity is illustrated
Figure 2.

Fig. 2. Proposed DAE to predict plant growth and productivity

B. Retailing of food refrigeration Depending upon the time in seconds, the variation among the
final defrost and observed values.
About 1000 refrigeration with real life datasets of 110000
train the data. To maintain food safety and quality, the created C. AWP based quality control in packaging and supply
Nemesyst model contains the ability to predict how much time chain
taken to turn off the refrigeration [11]. Whilst ascertain the The Adaptive Waterwheel Plant (AWP) algorithm offers
temperature of food never go through thresholds, the DAE several advantages, including efficiency, adaptability,
with refrigeration system is optimized. The cold energy robustness, convergence, parallelism, scalability, versatility,
storage and thermal inertia at such systems. Based on low sensitivity to initial parameters, potential for
refrigerator systems, the distinct multiple data with Nemesyst hybridization, and interpretability. These qualities make it a
model wrangles simultaneously. While consumers shops the valuable tool for solving a wide range of optimization
food in which the refrigeration retailing is changes based on problems in various fields. For the quality control while
thermal inertias. packing the foods for retails, it is ineluctable to provide
In an individual cases, the thermal inertia is determined optimized approach and for that we propose AWP algorithm.
that contributes to decarbonisation and ensuring food safety. While conducting the supply chain, the food safety is the
The refrigerator systems in each local shops interacted latter. important factor. To achieve this, the expiry date must be
From global database layer, a best network is retrieved in labeled exactly to avoid the poisoning of food. This will
which the local instance database is stored. The time till prevent the loss in economic and increase the productive [12-
temperature refrigerator increase after the points switched off 18]. The algorithm used here is elucidated below. The
is predicted. From the whole population, select the fridge waterwheel plant traps the wide petiole bears and act as the
subsets and the energy minimization is required to balance transparent flytraps. The possible solution of the universe and
cool. The stock affected potentially and the refrigerator is its respective vector issues are matrix in the following eqn.
affected to minimize the sufficient power consumption.

Authorized licensed use limited to: University of the Cumberlands. Downloaded on July 03,2024 at 19:26:11 UTC from IEEE Xplore. Restrictions apply.
 color. The actual yield prediction value is higher than all the
 R1   R1,1 L R1, k L R1, n 
other approaches and our proposed approach is more or less
M   M O M N M

   
(2)
similar to the actual value. The other approaches such as VMI
R =  RT  =  RT ,1 L RT , k L RT , n  [6], FSCM [7], and SCLIC [8] achieved lower prediction
    value and are visualized in figure. For the yield prediction of
M  M N M O M  plant we go with our proposed work.
 R J   R L R J ,T L R J , I 
 J ,1 
The number of variables and waterwheels are I and J. the
location of the waterwheels population is R with respect to the
packing of the food. The finalized weight is described as,
R1,T = DG k + Rand 1,k ⋅ (DQ k − DG k )
(3)
where , T = 1, 2,..., M and k = 1, 2,..., n
The lower and upper limits of the solutions are DG k and
DQk with the kth variables. The candidate solutions of
waterwheels are in the kth dimension.
D. Identify position and insects hunting
To find the new waterwheel position towards the insects is
expressed as,
Fig. 3. Performance evaluation based plant yield prediction over the data

(
J = Rand 1 ⋅ P (t ) + 2 s ) (4)
points

The performance evaluation based on expiry data

P (t + 1) = P (t ) + J ⋅ 2 s + Rand 2( ) (5) prediction rate is outlined in figure 4. Further, quality control
is an important factor and for that analysing or labelling the
The location of the waterwheel is varied and based on the exact expiry date is essential. This our work we checked with
Taylor series, the updated locations of the waterwheels are, the expiry data and for that we used the parameter expiry date
prediction rate and compared with the existing approaches
P ` ( k ) P `` ( k )  R ( t ) + 2i  (6) such as SCM [5], VMI [6], FSCM [7], and SCLIC [8]. The
P (t + 1) = P ( j ) + + + Rand 2  
1! 2!  J  prediction rate of our proposed approach is higher with
 
97.68% and other approaches such as SCM [5], VMI [6],
The circle diameter of the diameter is J with the old and FSCM [7], and SCLIC [8] accomplished the prediction rate of
`` 90%, 89%, 92%, and 94% respectively.
current solution as P (k ) and P ( k ) .
E. Insect carrying
In the exploitation stage the location shifts with the
maximized objective function and is framed as,

(
J = Rand 3 ⋅ i P best (t ) + Rand 3 P(t ) ) (7)

P (t + 1) = P (t ) + i J (8)

The best solution is with the current solution is P best and

P.
IV. EXPERIMENTAL INVESTIGATION AND DISCUSSION
The validation of proposed approach is elucidated in the
following section. In this the comparative study of the
proposed work with existing works are discussed. For the Fig. 4. Performance evaluation based on expiry data prediction rate
simulation we have taken the Python libraries with the
association of system with the specifications of Intel core i7 Figure 5 illustrates the false alarm rate prediction to save
11800H CPU, 16 GB RAM, RTX 3060 graphics card. The the time and also avoid spoiling the good foods. The false
investigations of results are discussed below. alarm rate of proposed is low and thus saves the time and
avoids food damaging with the percentage of 0.4%. The other
A. Performance Evaluation approaches such as SCM [5], VMI [6], FSCM [7], and SCLIC
The plant yield prediction by various approaches such as [8] achieved false alarm arte of 5.67%, 6.43%, 4.32%, and
proposed and VMI [6], FSCM [7], and SCLIC [8] are 3.15% respectively.
discussed in figure 3. The prediction of yield is measured in
kg/m2 over the data point in weeks. The actual prediction
value is marked in green and the proposed is spotted in red in

Authorized licensed use limited to: University of the Cumberlands. Downloaded on July 03,2024 at 19:26:11 UTC from IEEE Xplore. Restrictions apply.
 Fig. 7. Performance evaluation based on accuracy rates

Deep learning offers significant potential to enhance

Fig. 5. Performance evaluation based on false alarm rate
various aspects of food supply chain management, including
demand forecasting, quality control, supply chain
Figure 6 illustrates the false alarm rate prediction to save
optimization, predictive maintenance, personalized
the time and also avoid spoiling the good foods. The false
recommendations, supply chain visibility, risk management,
alarm rate of proposed is low and thus saves the time and
fraud detection, and continuous improvement. By leveraging
avoids food damaging with the percentage of 0.4%. The other
deep learning effectively, companies can improve efficiency,
approaches such as SCM [5], VMI [6], FSCM [7], and SCLIC
reduce costs, enhance food safety, and deliver value to
[8] achieved false alarm arte of 5.67%, 6.43%, 4.32%, and
customers.
3.15% respectively. Figure 5 illustrates the visualization of
missing detection and our proposed achieves the lowest with V. CONCLUSION
the 2.65%.
In order to anticipate plant productivity and an efficient
supply chain, this study developed a revolutionary deep auto-
encoder with the AWP algorithm, which minimizes both
environmental pollution and food waste. The implementation
of this effort, which gathers environmental and historical
growth data, is handled by MATLAB software. The ADE
model is expected to be trained using the yield and growth in
the time series. The large suppliers have influence over the
amount of energy used. Due to automatic control, the retail
refrigerator model temporarily shuts off during hours of peak
energy consumption. When the AWP algorithm is used to
automatically determine the retail packaged food expiration
date, food waste is decreased and consumer health and food
safety are guaranteed. Along with the plant yield prediction
based on the weeks' worth of data, the performance is
examined for the suggested and cutting-edge works, including
SCM, VMI, FSCM, and SCLIC, in terms of quality control
Fig. 6. Performance evaluation based on missing detection
metrics such expiration data prediction, missing detection, and
Figure 7 illustrates the accuracy rate prediction to save the false alert. In future, the authors would like to extent this
time and also avoid spoiling the good foods. The false alarm research in the topics such as Demand Forecasting and
rate of proposed is low and thus saves the time and avoids food Planning, Inventory Optimization, Quality Control and
damaging with the percentage of accuracy is 96.6%. The other Traceability, Sustainability and Waste Reduction, Supply
approaches such as SCM [5], VMI [6], FSCM [7], and SCLIC Chain Risk Management, Personalized Nutrition and Product
[8] achieved accuracy of 75%, 87%, 91% and 93% Recommendations, Humanitarian Logistics and Food Aid
respectively. Figure 5 illustrates the visualization of accuracy Distribution.
and our proposed achieves the highest with the 2.65%. REFERENCES
Integrating technology into supply chain management in the
[1] Wamba, S.F. and Queiroz, M.M., “Blockchain in the operations and
food industry offers numerous benefits, including improved supply chain management: Benefits, challenges and future research
visibility, enhanced efficiency, better decision-making, and opportunities”, International Journal of Information
greater sustainability. By leveraging advanced technologies, Management, vol. 52, p.102064, 2020.
companies can overcome challenges, optimize operations, and [2] Tang, X., “Research on smart logistics model based on Internet of
stay competitive in a rapidly evolving marketplace. Things technology”, IEEE Access, vol. 8, pp.151150-151159, 2020.
[3] Bukhari, A.H., Raja, M.A.Z., Sulaiman, M., Islam, S., Shoaib, M. and
Kumam, P., "Fractional neuro-sequential ARFIMA-LSTM for
financial market forecasting”, IEEE Access, vol. 8, pp.71326-71338,
2020.

Authorized licensed use limited to: University of the Cumberlands. Downloaded on July 03,2024 at 19:26:11 UTC from IEEE Xplore. Restrictions apply.
[4] Amrr, S.M., Asghar, M.J., Ashraf, I. and Meraj, M., “A comprehensive [12] Abdelhamid, A.A., Towfek, S.K., Khodadadi, N., Alhussan, A.A.,
review of power flow controllers in interconnected power system Khafaga, D.S., Eid, M.M. and Ibrahim, A., “Waterwheel Plant
networks”, IEEE Access, vol. 8, pp.18036-18063, 2020. Algorithm: A Novel Metaheuristic Optimization Method”, Processes,
[5] Bhutta, M.N.M. and Ahmad, M., “Secure identification, traceability vol. 11, no. 5, p.1502, 2021.
and real-time tracking of agricultural food supply during transportation [13] Kang, Z., Zhao, Y., Chen, L., Guo, Y., Mu, Q. and Wang, S.,
using internet of things”, IEEE Access, vol. 9, pp.65660-65675, 2022. “Advances in machine learning and hyperspectral imaging in the food
[6] Omar, I.A., Jayaraman, R., Salah, K., Debe, M. and Omar, M., supply chain”, Food Engineering Reviews, vol. 14, no. 4, pp.596-616,
“Enhancing vendor managed inventory supply chain operations using 2022.
blockchain smart contracts”, IEEE Access, vol. 8, pp.182704-182719, [14] Sharma, R., Kamble, S.S., Gunasekaran, A., Kumar, V. and Kumar, A.,
2020. “A systematic literature review on machine learning applications for
[7] Peña, M., Llivisaca, J. and Siguenza-Guzman, L., “Blockchain and its sustainable agriculture supply chain performance”, Computers &
potential applications in food supply chain management in Ecuador”, Operations Research, vol. 119, p.104926, 2020.
in Advances in Emerging Trends and Technologies, vol. 1, pp. 101- [15] Shahbazi, Z. and Byun, Y.C., “A procedure for tracing supply chains
112, 2020. for perishable food based on blockchain, machine learning and fuzzy
[8] Nagarajan, S.M., Deverajan, G.G., Chatterjee, P., Alnumay, W. and logic”, Electronics, vol. 10, no. 1, p.41, 2022.
Muthukumaran, V., “Integration of IoT based routing process for food [16] Kilimci, Z.H., Akyuz, A.O., Uysal, M., Akyokus, S., Uysal, M.O., Atak
supply chain management in sustainable smart cities”, Sustainable Bulbul, B. and Ekmis, M.A., “An improved demand forecasting model
Cities and Society, vol. 76, p.103448, 2022. using deep learning approach and proposed decision integration
[9] Zhiwen, Z., Yujun, X., Junxing, L., Limin, G. and Long, W., “Supply strategy for supply chain”, Complexity, 2019.
chain logistics information collaboration strategy based on [17] Punia, S., Singh, S.P. and Madaan, J.K., “A cross-temporal hierarchical
evolutionary game theory”, IEEE Access, vol. 8, pp.46102-46120, framework and deep learning for supply chain forecasting”, Computers
2020. & Industrial Engineering, vol. 149, p.106796, 2020.
[10] Xie, R., Wen, J., Quitadamo, A., Cheng, J. and Shi, X., “A deep auto- [18] Husna, A., Amin, S.H. and Shah, B., “Demand forecasting in supply
encoder model for gene expression prediction”, BMC Genomics, vol. chain management using different deep learning methods” in Demand
18, pp.39-49, 2020. forecasting and order planning in supply chains and humanitarian
[11] Nemesyst. https://github.com/DreamingRaven/Nemesyst. logistics, pp. 140-170, 2021.

Authorized licensed use limited to: University of the Cumberlands. Downloaded on July 03,2024 at 19:26:11 UTC from IEEE Xplore. Restrictions apply.

View publication stats