Page 87 - Kaleidoscope Academic Conference Proceedings 2024

Innovation and Digital Transformation for a Sustainable World

Table 2 – Standardization Progress of C-V2X Sidelink

V2X Sidelink Standardization Milestones
LTE-V2X sidelink was included in the 3GPP standardization for Release 14 (R14) and
LTE-V2X (R14 and r15)
Release 15 (R15). R14, finalized in March 2017, aimed to support fundamental road
safety applications using LTE sidelink [17].
R15, completed in June 2018, further enhanced the support for advanced V2X applications
utilizing LTE sidelink technology. It introduced the concept of direct vehicle-to-vehicle
communication, known as PC5, to facilitate V2X interactions [18].
5G-V2X sidelink was introduced in Release 16 (R16), completed in June 2020. R16
brought enhancements to V2X applications using 5G sidelink, providing lower latency,
5G-V2X (R16, R17, and R18)
higher reliability, and increased data rates for V2X communications [19], [20].
Release 17 (R17), aimed at further advancing V2X applications, includes additional
improvements to the 5G-V2X sidelink. R17 focuses on enhancing performance for
cooperative and coordinated driving scenarios, enabling more complex V2X use cases
[21].
Release 18 (R18), expected to be finalized by the end of December 2023, will continue
the evolution of sidelink technology in 5G-V2X. R18 is anticipated to support even more
challenging V2X scenarios, such as high-density environments and high-speed mobility,
further improving the efficiency and reliability of V2X communications [22].
With ongoing standardization efforts, 5G-V2X sidelink is set to provide comprehensive
support for a wide range of V2X applications, driving the future of connected and
autonomous vehicles [22].

Mathematical Model: 4.3.2 Machine Learning for Predictive Resource Allocation

Building on the insights from [3, 6], we integrate machine
learning techniques to predict and adapt to traffic conditions
∑︁ ∑︁
∑︁
min , ( , ) + ( ) (1) dynamically, enhancing the real-time operational capacity of
,
=1 =1 =1 sidelink V2X communications.
Algorithm 2 Deep Learning-Based Predictive Resource
∑︁
s.t. , = , ∀ ∈ {1, . . . , }, (2)
Allocation
=1
1: Data: Traffic data, resource usage patterns
0 ≤ , ≤ 1, ∀ ∈ {1, . . . , }, ∀ ∈ {1, . . . , } (3)
2: Result: Predictive allocation model
3: Initialization: Train a deep neural network on historical
data
where , is the proportion of spectrum allocated to vehicle 4: while operational do
5: Collect real-time traffic and resource usage data
at time , is the total available spectrum at time , ,
represents the utility function of allocation for vehicle at time 6: Predict future resource requirements using the trained
, and is a penalty function enforcing smooth variations model
in total spectrum allocation, with as a regularization 7: Allocate resources according to the predictions to
parameter. optimize network efficiency
Algorithm 1 Enhanced Spectrum Allocation for Sidelink 8: Retrain the model periodically with new data
V2X Communication 9: end while
1: Data: Utility functions , , penalty function , total
We also introduce a machine learning-based optimization
spectrum , regularization parameter
method for resource allocation which uses reinforcement
2: Result: Allocation matrix ,
learning to dynamically adjust resource allocation based on
3: Initialization: Allocate the initial spectrum based on
predicted traffic conditions.
historical data
4: for each time step do
5: Evaluate current network conditions and vehicle ∑︁
demands L( ) = − log ( | ), (4)
6: Solve the optimization model to update , =0
7: Implement spectrum allocation
8: Adjust future allocations based on observed where L( ) is the loss function, is the policy
performance parameterized by , is the action taken at time , is
9: end for the state at time , and is the reward at time .

– 43 –