Parametric evaluation of highway redesign models for traffic

flow optimization from a statistical perspective

D Thakur1, P Wadhai2, B Khode3
1 Post
Graduate Scholar Transportation Engineering, G. H. Raisoni College of
Engineering, Nagpur, India.
2Head of Department, Civil Engineering, G H Raisoni College of Engineering, Nagpur,
India 3Associate Professor, Civil Engineering, G H Raisoni College of Engineering, Nagpur,
India E-mail: dipannita.thakur.mtechtrs@ghrce.raisoni.net

Abstract. Highway redesign is a multidomain process that involves an analysis of road

density, identification of local road networks, estimation of critical junctions, study of the
ratio of left to right turns, identification of directional lanes, storage space compu tation,
and other related subjects. These characteristics must be examined in order to determine
cycle durations, flow ratios, and other critical parameters. Researchers have suggested a
broad range of such models, each with its own set of internal operational properties.
Models that perform these tasks differ in terms of their qualitative and quantitative
performance, making model selection for real-time deployment situations more difficult.
This paper examines several highway redesign suggestion models in terms of their
cumulative subtleties, contextual benefits, application-specific limits, and functional future
research areas in order to eliminate ambiguity. Researchers will be able to find
performance-specific models for their needs, reducing the amount of uncertainty in the
model selection process. Based on the results of this debate, it was discovered that current
models that use machine learning recommendation approaches beat other models in terms
of suggestion efficiency. Estimations of recommendation accuracy, computational
complexity, deployment cost, scalability, and accident-avoidance capabilities were used to
assess this performance. Researchers will be able to find the best models for performance-
aware roadway redesign suggestion applications based on this analysis.

Keywords: Highway, Redesign, Recommendation, Machine, Learning, Accident, Traffic

1. Introduction
Highway redesign recommendation is a multidomain process that involves evaluation of road
density, identification of local road networks, estimation of critical intersections, evaluation of
left & right turning ratio, identification of directional lanes, calculation of storage areas, etc.
Analysis of these parameters is used for calculation of cycle lengths, flow ratio, and other
 critical parameters. A typical model flow for that is used for highway redesign
recommendations can be observed from figure 1, wherein estimation of these parameters along
with recommendation of new lanes, intersections, etc. is discussed for different highway types
[1]. This model is capable of improving recommendation efficiency via estimation of grade
separation, pick up points, inflow & outflow points, directional lanes, phasing plans, etc. The
model evaluates a flow variance via equation 1, which assists in identification of traffic flows
for each intersection, and uses flow (f), and their average levels (f¯) for N different iterations
(f),
Σ(𝑓 − f¯)2
𝑣=√
𝑁

The model aims at minimizing variance between different recommendations, which assists in
formation of a better highway that can provide consistent traffic flows.

Figure 1. A typical highway recommendation model for multiple scenarios

Similar models are proposed by researchers, and their cumulative nuances, contextual
advantages, application-specific limitations, and functional future research scopes are
discussed in the next section of this text [2, 3, 4, 5, 6]. Based on this discussion, it was
observed that machine learning techniques outperform linear recommendation techniques, and
thus must be used for real-time highway designs. Section 3 compares these techniques in terms
of recommendation accuracy, computational complexity, deployment cost, scalability, and
accident-avoidance capabilities, which will assist readers to identify performance-specific
models for their use cases. Finally, this text is concluded with various performance specific &
context specific observations about reviewed models, and also recommends methods to further
optimize their performance levels.
2. Literature Review
For thruway redesign recommendations, various models are introduced, each with its own
arrangement of inward working highlights. The region encompassing the intersection
uniqueness is depicted by [1] as a bottleneck for the thruway. On the off chance that a two-path
exit can't deal with how much traffic, the motorway's viability will diminish. To help the limit
of the uniqueness region, organizers recommended constructing a three-path exit. Existing
Chinese plan necessities don't consider a three-path exit, and growing slope paths by building
helper paths as per conventional plan standards would protract the mainline broadening
segment, expanding both development land utilization and venture costs. This study
recommends that a two-path exit (TLE) in Xi'an be extended to a three-path exit. With the
utilization of VISSIM, three very small reenactment models of wandering zones were made.
The loads of four appraisal markers were impartially resolved utilizing entropy, and a full
assessment and application investigation of the ongoing circumstance, as well as plans for a
three-path exit, was directed. As per [2,] the cost court's ability fundamentally affects the
street's general help quality. The limit of cost squares during top periods is connected to the
comparing functional costs. In this examination, it is expected that in a serious transportation
framework, the local location and working environment are associated by a parkway with a
cost square that runs lined up with a mass travel line, with suburbanites having the choice of
driving or utilizing public transportation. The objective of this examination is to perceive what
such a framework means for gridlock and air contamination. Cost court limit plans were
examined all through the traffic administration period to fulfill three objectives: cost square
earn back the original investment, benefit expansion, and by and large cultural transportation
cost decrease. This was achieved by means of the improvement of a balance model and two
numerical programming models (EMP). Three unmistakable outing balance conditions were
utilized to concentrate on the modular parts. Furthermore, a bi-objective streamlining model
fully intent on expanding both all out benefit and complete social expense was planned and
surveyed. The hypothetical examination was upheld by mathematical models.

In light of the previously mentioned information, work [3] assessed genuine traffic lull and cut-
in circumstances across German and Japanese expressway direction datasets. The two
situations were made utilizing traffic information assembled by instrumented cars and fixed
cameras along motorways in Japan and Germany (the SAKURA dataset) (highD dataset). Five
separate vehicle kinematic factors were utilized to address the two circumstances, including
sidelong and longitudinal distances, speeds, and speed increases. The relationship and crossing
point objective estimations, as well as security measurements like opportunity to-crash and
time-progress, were utilized to analyze these factors. In spite of varieties in rush hour gridlock
speed limitations, street plan, and information sources between the two countries, the
aftereffects of the information examination uncovered significant relationships and
convergences of boundary circulation for the two situations. The two circumstances have a
Time-to-Collision that is like one another. Contrasts in time progress uncover that the protected
distance varies between the two countries, suggesting that security evaluation strategies should
be custom fitted to each putting's one-of-a-kind qualities together to guarantee individuals'
wellbeing. These discoveries show that overall wellbeing pointers might be created, and they
propose further examination into information gathering and correlation to deliver normalized,
comprehensively material, and area nonpartisan methodologies for surveying the security of
independent vehicles (AVs). Charge and hybridization of rough terrain vehicles, as per [4,]
could save fuel and lessen how much contamination made. Due to the high power and power
thickness necessities, as well as the different driving cycles, energizing these frameworks is a
troublesome activity. This examination explores a clever method for stimulating rough terrain
vehicles using a direct cylinder siphon and a straight actuator. This piece of the paper starts
with a conversation of straight actuator machine geographies for charge siphon applications. A
rounded machine model is being worked determined to incorporate it with a pressure driven
siphon. For recreations of the position controllers and the entire charge siphon plan, a
Multiphysics system (MPF) is utilized.

The point mists gathered by a vehicle-borne portable laser checking (MLS) framework have a
ton of potential for applications like smart transportation frameworks, street foundation stock,
and superior quality (HD) maps, which will help independent vehicles and high level driver
help frameworks (ADAS) work all the more securely (AVs). The utilization of MLS direct
mists in a two-step strategy toward mechanized parkway bend acknowledgment and remaking
is recommended in this paper. While managing loud, unstructured, thick point mists, specialists
defy issues, for example, controlling exceptions created during the street marker discovery
process and perceiving the many sorts of bends made during 3D bend recreation. The two
systems utilized in the development of our answer are a model fitting assessor and a power
fluctuation finder. Tests using both a virtual sweep dataset and a genuine MLS dataset
uncovered that our technique has a ton of commitment for controlling exceptions and
producing 3D street twists. The circle radii registered from the virtual output dataset are exact
to inside 0.6 percent. As per the consequences of a new exploration, our innovation for
distinguishing street signs is more reliable and exact than other state of the art frameworks. As
per reference [6, platooning associated and independent vehicles (CAVs) increments
throughput. Development of companies may be trying because of communications among
CAVs and different vehicles. Scientists present another liquid model of blended independence
traffic stream to investigate and make detachment coordination strategies. This model is
accessible here. The scientists furnish a couple connect liquid model with haphazardly showing
up companies and other non-CAV vehicles. Specialists had the option to lay out a Foster-
Lyapunov float condition for the liquid model by assessing a M/D/1 lining instrument. In light
of these reliable qualities, doing quantitative assessments of parkway throughput is simple. The
detachment's throughput is improved while inertness is diminished because of the
methodology. These liquid model planning systems have been demonstrated to be successful
and easy to utilize. Our discoveries were affirmed utilizing both infinitesimal and perceptible
reenactment models (cell transmission model and metropolitan versatility model, individually).

An ascent in the use of shrewd vehicles and vehicular impromptu organizations (VANETs), as
per [7], has brought about a larger number of people having Internet access. Helpful
downloading through drive-through Internet is one of the as of now open other options, despite
the fact that it has various downsides. Picking reliable intermediary cars in a high-traffic
driving climate and accomplishing a harmony between adaptable information downloading and
satisfactory security are two of these issues. Scientists utilize a choice methodology in light of
a client's standing as well as requested marks to create a dependable, versatile, and versatile
helpful downloading component. Just intermediary vehicles with the best projected
downloading limit will be used while picking intermediary vehicles. This empowers us to pick
solid intermediary vehicles. Adaptable information downloading involves separating a
document into incredibly little pieces so an intermediary vehicle might change the download in
light of a few circumstances, like the ongoing climate. Our method likewise gives a serious
level of safety, since it utilizes traditional confirmation, safeguards clients' security, and keeps
interchanges stowed away (i.e., process validation: verifies the request for record blocks and
the request for the helping vehicles utilizing requested marks). Expressway burrows have a
higher risk of mishap than open thruways, requiring a deliberate way to deal with guarantee
burrow wellbeing. A few hardships were found through past examination: 1) Tunnels haven't
stood out enough to be noticed as open interstates.3) None of the examinations checked out at
the connection between unseen heterogeneity and abundance zero or zero perceptions. To
address these challenges, the creators of this paper originally made a reasonable dataset for
burrows. This dataset contains impact information from various Chinese roadways and
passages throughout the span of three to five years, as well as data with respect to their plan
highlights, traffic conditions, and asphalt conditions. The associated irregular boundaries
negative binomial Lindley (CRPNB-L) model was proposed, which represents both abundance
zero perceptions and unseen heterogeneity. The CRPNB-L model consolidates the Lindley
dissemination to represent unnoticed heterogeneity and to address the issue of dissimilarity
from abundance zero information. Likewise, the model thinks about the related irregular
boundaries. The CRPNB-L model has the best decency of-fit in examinations. Fragment
length, traffic volume, level of class 5 vehicles (huge trucks and trailers), burrow passage and
leave sections, and steep uphill and downhill portions were undeniably demonstrated to be
emphatically associated with crash recurrence. Then again, boundaries including shape, burrow
length, asphalt harm condition record (PCI), and slip opposition file were demonstrated to be
antagonistically associated with mishap recurrence (SRI). Ebb and flow, a steepness file, the
level of class 5 vehicles, and the SRI were completely recognized and connected.

Work in [9] is an examination project held back nothing plan and execution for dubious
mechanical frameworks. As a rule, the presence of vulnerability in an actual framework is
unavoidable, and the framework's potential cutoff decides this. Take, for example, non-straight
mechanical frameworks. Indeed, even the bound coefficients are a secret, and it's hazy the way
that they're connected to a specific set. The exhibition of a progressive framework is required.
The main level is described by determinism, homogeneity, and boundedness in its definitive
structure. In spite of the absence of lucidity, this part should be finished completely. The
subsequent stage involves however much practical decrease of a fluffy hypothetical exhibition
record (FTPI). The scientists have conceived a control conspire that is versatile. Inside the
provided boundaries, the control gives the most ideal exhibition. Analysts have consolidated
range adaptability to meet the rules of the subsequent level. Taking care of a streamlining issue
to get the best incentive for a plan boundary is required. The issue has been settled. The most
minimal expense and insightful plans of the plan boundary are given by specialists. Specialists
had the option to make a two-level control plan. The Vehicular Network (VANET) [10], which
vows to improve both driving wellbeing and outing solace, has provoked the consideration of
specialists in scholarly community and business from one side of the planet to the other. The
VANET might have the option to improve both street wellbeing and travel solace by utilizing
vehicle-to-vehicle (V2V) and vehicle-to- framework (V2I) joins. This would be finished by
expanding drivers' attention to their environmental factors and providing travelers with
amusement related data. The inherently liquid nature of VANET's organization configuration
makes information move and content conveyance difficult to guarantee. Specialists present a
novel information parkway helped content conveyance framework for VANETs in this
distribution. Specialists take a gander at the advantages of foreseeing vehicular traffic volume
utilizing a calculation that can rapidly gauge traffic stream in this convention. Specialists
utilize three distinct informational indexes from England Highways to test the proposed traffic
stream forecast technique. A lot of reproduction has gone into the recommended approach for
material dispersion.

Eco-accommodating preservationists are worried about the amount of fuel consumed via
automobiles and the discharges of gases, as per [11]. Because of different specialized upgrades,
the typical mileage of cars that are driven consistently has developed. There are a few
techniques for lessening the amount of fuel utilized and contamination made via vehicles. The
majority of these ideas were made for urban areas and metropolitan districts, which utilize
substantially more energy. Roadway clients cover a lot of ground. Gas utilization might ascend
as a result of somewhat little mistakes; however, when drivers go with similar wasteful choices
more than once, the size of the expansion in overabundance fuel utilization might detonate.
Specialists need to fabricate a green convention as a component of this program, with the goal
of supporting drivers and self-driving vehicles in bringing down their dependence on
petroleum derivatives and the contamination they cause. The objective of this system is to keep
internal combustion vehicles at a consistent speed to save energy and increment mileage. It
works on the consistency of the vehicle's speed increase and deceleration. A total series of
trials was utilized to survey the presentation of the proposed strategy. A count of the on-street
vehicles by and by present on the course is basic for effective transportation the board [12].
This paper proposes the principal approach for counting cars on streets that works in a packed
space while as yet keeping up with pixel-space exactness. Since there is restricted data on the
vehicles and a ton of precariousness in the compacted space, counting is testing. To move
beyond the restricted video information, scientists build low-level qualities. Mining coding-
related information to distinguish new attributes is conceivable. The specialists then propose a
HCR model that uses compacted area low- level qualities for each casing to foresee the
quantity of vehicles. To portray the wide assortment of such conditions, HCR progressively
bunches traffic scenes concurring on the quantity of the vehicles present. Scientists recommend
utilizing a limited worldly relapse model, which considers the ceaseless change in rush hour
gridlock stream, to further develop the counting discoveries. Video from parkway surveillance
cameras is utilized to survey the arrangement. The aftereffects of the tests recommend that the
proposed technique can rival pixel-based arrangements. These arrangements earn a similar
college education of execution at a fundamentally diminished cost of figuring.

Right evening out is used to check the lopsided deformity of the expressway span head, as per
the review given in [13], which is both unsafe and tedious to attempt. This work presents a
strategy for getting parkway span head deformity properties from TLS point mists. A robotized
information assortment framework is utilized to figure the distance between filtering stations
and their goal. The plane fitting method is utilized to denoise street point mists and gauge the
trial limit. The unrefined point cloud enlistment depends on data from street partitioning lines,
while the refined point cloud enrollment depends on the weighted normal height. The level of
the disfigurement checking point is the normal of the multitude of focuses in the predefined
framework surface, which lessens arbitrary rise errors. The viability of the proposed technique
was tried on three unique courses. The height check precision is superior to 2 millimeters, and
the rise repeatability is more noteworthy than 1 millimeter, as per the information,
demonstrating that the deformity observing boundaries have been fulfilled. At a bridgehead
station, information gathering requires three to four minutes. The depicted versatile TLS-based
approach might be utilized to research interstate scaffold head distortion. Thus, [14] mirrors the
decisions that a driver would make while converging at an expressway convergence. Then, to
integrate conduct control, demonstrating is utilized. Initial, a driving test system is utilized to
look at driving conduct in the essential path at a roadway convergence. The driver's condition
of choice (SOD), which shows endorsement for blending a vehicle from the consolidating path
into the primary path, is offered unique consideration. Second, the SOD of the driver is
displayed utilizing strategic relapse, and the model's expectation capacity is assessed. The
speed regulator on the consolidating vehicle is intended to make driving in the principal path
more endurable. By adjusting the speed of the consolidating vehicle, the entropy of the driver
of the auto in the fundamental path might be diminished. A dynamic model is utilized to
quantitatively measure the issue, and streamlining is performed by means of a randomized
procedure. Therefore, self-driving vehicles may now proceed onto streets in a purposeful way.
Mathematical testing is utilized to show the viability of the plan thoughts.

Most of Earth perception research [15] centers around making calculations for removing
semantic data from VHR flying photos. Infrared pictures, optical pictures, LiDAR point mists,
and information from inertial estimation units (IMUs) are only a couple of instances of the
multimodal data that every sensor might give. LiDAR and VHR profound organizations don't
make full benefit of multimodal information in their current cycles. The primary changes that
happen across modalities, as well as the measurable highlights that are grown physically, are
not thought about by MFNet. Specialists made another profound MFNet that can utilize
multimodal VHR aeronautical photographs and LiDAR information, as well as the intramodal
properties that go with them. LiDAR-inferred highlights (like slant and standardized advanced
surface model (NDSM)) as well as symbolism determined highlights are incorporated (like
IRRG, NDVI, and DoG). From a specialized perspective, scientists use consideration and
multimodal figuring out how to incorporate both multi-purpose and intramodal viewpoints. A
multimodal combination strategy, pyramid widening blocks, and a multi-facet highlight
combination module were completely picked. Our organization had the option to accomplish
versatile multimodal information combination, expand the responsive field, and work on how
much worldwide to-nearby logical combination. The organization was improved with the
assistance of multiscale oversight preparing. Broad trial discoveries and removal testing
utilizing ISPRS semantic and IEEE GRSS DFC Zeebrugge datasets exhibit the adequacy of
MFNet. As indicated by [16], the compelling reception of connected independent driving
requires the utilization of organization based localisation. Accuracy and consistency under one
meter are fundamental for these applications. Street side units of the fifth era (otherwise called
RSUs) and vehicular organizations are vital for exact vehicle localisation. These organizations
are not intended to work in a given area; rather, they are intended to ship information, requiring
a novel sending and working technique. The exact vehicle-to-framework area might be laid out
involving the plan norms for devoted network establishments given in this article. To start, the
organization's design or the site's position is estimated mathematically. The base RSU
thickness all through the course is resolved utilizing likelihood models that record for view
conditions in metropolitan road and expressway circumstances. By utilizing a few radio wire
establishments, the Cramér-Rao approach for consolidated ToA and AoA limitation limits the
distance between RSUs along the course. This guarantees that area assessments are pretty
much as precise as could be expected. As per the recreation's discoveries, network destinations
ought to be put on inverse roadsides, with a most extreme distance between RSUs of 40 or 230
meters, it is metropolitan or provincial to rely upon whether the climate. Radio wire clusters
should incorporate uplink ToA and AoA estimations to satisfy the prerequisites of these plan
particulars. In 95% of cases, this will empower them to accomplish an area exactness of short
of what one meter. For each eight recieving wire components present at RSUs, the expressway
limitation network thickness is brought down by an element of 1.5.

The reasonability of driver-mechanization shared driving control in thruway traffic was

analyzed in [17], with the point of path keeping up with and impediment aversion. A unique
helpful direction arranging strategy and a fluffy guiding regulator make up the common control
engineering. To create helpful direction arranging, polynomial capabilities are used. Driver
observing, question settlement, and move determination are factors in this preparation. The
vehicle's expected course modifies to think about the driver's activities and goals. Utilizing
Lyapunov soundness contemplations, scientists might have the option to build a Takagi-
Sugeno Fuzzy (TS Fuzzy) model-based regulator. This guiding regulator's help force makes it
simple for the vehicle to head down the ideal path. The new common driving control design
gives drivers more path turning opportunity while at the same time bringing down how much
struggles between mechanized frameworks and drivers. The outcome of the exhorted approach
is surveyed utilizing a dynamic, intuitive driving test system, as well as goal and emotional
information from genuine drivers. The work in [18] fabricates a nonlinear spectator that is
utilized in mix with a radar sensor to follow moving vehicles' courses. In past examination on
vehicle following, a different model channel (TVMF) was used, which contained many models
to represent the different movement examples of vehicles. In this work, a nonlinear vehicle
model is utilized to depict the vehicle's all's possible movement modes. On account of the
tremendous assortment of possible working conditions, the consequences of earlier
examination on nonlinear spectator configuration don't make a difference to the framework
under request. The interrelated nonlinear capabilities that support movement are exposed to
more grounded limits thanks to another way for developing nonlinear spectators. Lyapunov
calculations might be utilized to decide the spectator's dramatic strength. The two recreations
and testing utilize the recently evolved onlooker plan. The signs from the radar might be
utilized to decide the vehicle's direction, as well as its longitudinal, sidelong, and speed. The
discoveries are presented here to screen vehicle developments on thruways and at metropolitan
intersection with turns and significant bearing changes.

As per [19], the advances of vehicles and remote correspondences should be immovably
entwined to develop ITS organizations. The terahertz (THz) recurrence range, which has an
enormous transfer speed, may be utilized in remote correspondence to empower high-
information rate remote transmission past fifth-age portable correspondences (B5G). This
study portrays V2I channels running at 300 GHz with a 8 GHz transfer speed for use in
metropolitan and parkway conditions. A RT test system with affirmed EM settings is utilized
by the specialists. Analysts are taking a gander at the impact of precipitation on channel
portrayal when the transporter recurrence frequency moves toward the measures of snowflakes
and drops. Way misfortune, Rician K-factor, root-mean-square (RMS) postpone spread, and
point spreads are explored under radiant, stormy, and cold circumstances. The divert
boundaries portrayed in this study may be utilized in the structure of future THz vehicular
associations and frameworks. In agreeable versatile journey control, otherwise called CACC,
model prescient control, otherwise called MPC [20], has tracked down significant execution.
MPC's better exhibition as far as advancing mileage and adjusting to between vehicle distance
limits is the explanation for this. Since a MPC-based strategy is used to expect the way of
behaving of a vehicle utilizing a unique model, the space vehicle dynamic model and the time-
space vehicle dynamic model are in many cases utilized pair in ebb and flow research. This
article looks into the space-area and time-area MPC vehicle following models. The exhibition
of the two procedures is thought about and assessed utilizing a MPC configuration process and
scientific technique for rapid vehicle following. To investigate speed following and fuel-saving
execution, which are vehicle following objectives that rival each other, a bi-objective expense
capability has been created and weighted. The time-area approach outflanks the space-area
methodology with regards to efficiency. The utilization of room space models that contain
street information, as opposed to time-area models that just incorporate restricted street
information, works on both following and efficiency. A near Pareto examination was utilized
to look at the levels of following incorrectness, efficiency, and street level data. [21] researches
the subject of range partaking with regards to a one-way interstate vehicle organization.
Scientists allude to the V2V, V2I, and RSU hubs as 1D Poisson point processes on an
interminable straight line (PPPs). V2V hubs use an OSA method in view of conflict to reuse
V2I- involved range. In the event that it is fruitful, a V2V hub will vie for access authorization
with other qualified V2V hubs. A V2V hub's responsibility is to sort out where the holes in the
spatial range between V2I associations are. The transmission likelihood of V2V hubs is
impacted by this arrangement. Likewise, the inclusion of both V2V and V2I associations is
evaluated. The non- homogeneity of qualified/dynamic V2V hubs makes computing
transmission and inclusion likelihood essentially seriously testing. Specialists give a HPPP
estimate to the geographic model of qualified/dynamic V2V hubs and show that this guess is
effective through broad recreations. This makes it more straightforward to coordinate insightful
discoveries. Scientists embrace a review to recognize what the framework's qualities mean for
the organization's presentation to give vital plan bits of knowledge. Dispute based OSA might
have the option to further develop the range productivity of one-way interstate traffic
organizations, as per research in view of mathematical examination and reenactments.
Following this assessment, it was found that different models vary regarding inward functional
attributes. The following part delves further into these models' performance by examining their
cumulative subtleties, contextual benefits, application-specific limits, and functional future
research objectives, which will help readers identify these models in terms of their
performance-specific use cases.

3. Results & Performance Evaluation

Based on the review, it can be observed that Machine Learning Models (MLMs) outperform
others in terms of recommendation performance. Thus, in this section the reviewed models are
compared in terms of metrics including recommendation accuracy (A), computational
complexity (CC), deployment cost (DC), scalability (S), and accident-avoidance capabilities
(AAC), which are quantized into Low Value (LV), Medium Value (MV), High Value (HV),
and Very High Value (VHV) ranges. Based on this strategy, the comparison can be observed
from table 1 as follows,

Table 1. Performance analysis of different models

Model A AAC CC DC S
TLE [1] MV MV HV MV HV
EMP [2] MV LV LV HV HV
MPF [4] HV HV MV MV MV
ADAS [5] HV VHV MV HV VHV
CAVs [6] HV HV HV HV HV
CRPNBL [8] VHV HV HV HV HV
FTPI [9] HV HV HV HV HV
VANET [10] VHV HV HV VHV VHV
HCR [12] VHV VHV MV MV HV
TLS [13] HV HV HV HV HV
SOD [14] VHV HV HV MV HV
IMUs [15] MV HV MV LV HV
ToA & AoA [16] MV HV HV HV HV
TS Fuzzy [17] HV MV HV MV HV
TVMF [18] VHV HV MV HV VHV
MPC [20] VHV HV HV MV HV
 Figure 2. Comparative analysis of different models

Based on this analysis and figure 2, it can be observed that HCR [12], CRP NBL [8], VANET
[10], SOD [14], TVMF [18], and MPC [20] showcase higher accuracy of prediction, while
ADAS [5], and HCR [12] showcased better accident-avoidance capabilities, thus they can be
fused in order to obtain better highway recommendation performance. It can also be observed
that EMP [2], MPF [4], ADAS [5], HCR [12], IMUs [15], and TVMF [18] have lower
complexity, while IMUs [15], TLE [1], MPF [4], HCR [12], SOD [14], TS Fuzzy [17], and
MPC [20] have lower deployment costs, while ADAS [5], VANET [10], and TVMF [18]
showcase higher scalability, thus can be used for a wide variety of low cost, low complexity,
and high scalability applications.

4. Conclusion and Future Work

This paper reviews a wide variety of models for highway recommendation, and compares their
cumulative nuances, contextual advantages, application-specific limitations, and functional
future research scopes. Based on this comparison, it was observed that Machine Learning
Models (MLMs), and adhoc network models outperform other models, which makes them
useful for real- time applications. It was observed that HCR, CRP NBL, VANET, SOD,
TVMF, and MPC showcase higher accuracy of prediction, while ADAS, and HCR showcased
better accident-avoidance capabilities, thus they can be fused in order to obtain better highway
recommendation performance. It can also be observed that EMP, MPF, ADAS, HCR, IMUs,
and TVMF have lower complexity, while IMUs, TLE, MPF, HCR, SOD, TS Fuzzy, and MPC
have lower deployment costs, while ADAS, VANET, and TVMF showcase higher scalability,
thus can be used for a wide variety of low cost, low complexity, and high scalability
applications. Thus, these models and their combinations must be used for high-efficiency
deployments. In future, these models must be validated on larger scale applications, and can be
extended via fusion methods, which will assist in improving their applicability for large-scale
deployment scenarios.
5. References
[1] B. Pan, Z. Xie, S. Liu, Y. Shao and J. Cai, "Evaluating Designs of a Three-Lane Exit Ramp Based on the Entropy Method,"
in IEEE Access, vol. 9, pp. 53436-53451, 2021, doi: 10.1109/ACCESS.2021.3070465.
[2] G. Yong, J. Zhang, N. Su and X. Feng, "Capacity Design and Pareto Improvement of Highway Toll Plaza in a Competitive
Transport System," in IEEE Access, vol. 9, pp. 113642-113652, 2021, doi: 10.1109/ACCESS.2021.3104332.
[3] A. Zlocki et al., "Logical Scenarios Parameterization for Automated Vehicle Safety Assessment: Comparison of
Deceleration and Cut-In Scenarios From Japanese and German Highways," in IEEE Access, vol. 10, pp. 26817-26829,
2022, doi: 10.1109/ACCESS.2022.3154415.
[4] A. Khamitov, J. Swanson, J. Van de Ven and E. L. Severson, "Modeling, Design, and Testing of a Linear Electric-
Hydraulic Conversion Machine for Electrification of Off-Highway Vehicles," in IEEE Transactions on Industry
Applications, vol. 57, no. 3, pp. 2449-2459, May-June 2021, doi: 10.1109/TIA.2021.3066084.
[5] Z. Zhang, J. Li, Y. Guo, C. Yang and C. Wang, "3D Highway Curve Reconstruction from Mobile Laser Scanning Point
Clouds," in IEEE Transactions on Intelligent Transportation Systems, vol. 21, no. 11, pp. 4762-4772, Nov. 2020, doi:
10.1109/TITS.2019.2946259.
[6] L. Jin, M. Čičić, K. H. Johansson and S. Amin, "Analysis and Design of Vehicle Platooning Operations on Mixed-Traffic
Highways," in IEEE Transactions on Automatic Control, vol. 66, no. 10, pp. 4715-4730, Oct. 2021, doi:
10.1109/TAC.2020.3034871.
[7] Y. Zhang, L. Zhang, D. Ni, K. -K. R. Choo and B. Kang, "Secure, Robust and Flexible Cooperative Downloading Scheme
for Highway VANETs," in IEEE Access, vol. 9, pp. 5199-5211, 2021, doi: 10.1109/ACCESS.2020.3048273.
[8] F. Tang, X. Fu, M. Cai, Y. Lu, S. Zhong and C. Lu, "Applying a Correlated Random Parameters Negative Binomial
Lindley Model to Examine Crash Frequency Along Highway Tunnels in China," in IEEE Access, vol. 8, pp. 213473-
213488, 2020, doi: 10.1109/ACCESS.2020.3040211.
[9] R. Yu, Y. -H. Chen, B. Han and H. Zhao, "A Hierarchical Control Design Framework for Fuzzy Mechanical Systems With
High-Order Uncertainty Bound," in IEEE Transactions on Fuzzy Systems, vol. 29, no. 4, pp. 820-832, April 2021, doi:
10.1109/TFUZZ.2020.2965913.
[10] P. Sun, N. AlJeri and A. Boukerche, "DACON: A Novel Traffic Prediction and Data-Highway-Assisted Content Delivery
Protocol for Intelligent Vehicular Networks," in IEEE Transactions on Sustainable Computing, vol. 5, no. 4, pp. 501- 513,
1 Oct.-Dec. 2020, doi: 10.1109/TSUSC.2020.2971628.
[11] M. B. Younes and A. Boukerche, "Towards a Sustainable Highway Road-Based Driving Protocol for Connected and Self
Driving Vehicles," in IEEE Transactions on Sustainable Computing, vol. 7, no. 1, pp. 235-247, 1 Jan.-March 2022, doi:
10.1109/TSUSC.2021.3074596z
[12] Z. Wang, X. Liu, J. Feng, J. Yang and H. Xi, "Compressed-Domain Highway Vehicle Counting by Spatial and Temporal
Regression," in IEEE Transactions on Circuits and Systems for Video Technology, vol. 29, no. 1, pp. 263-274, Jan. 2019,
doi: 10.1109/TCSVT.2017.2761992.
[13] M. Liu, X. Sun, Y. Wang, Y. Shao and Y. You, "Deformation Measurement of Highway Bridge Head Based on Mobile
TLS Data," in IEEE Access, vol. 8, pp. 85605-85615, 2020, doi: 10.1109/ACCESS.2020.2992590.
[14] H. Okuda, T. Suzuki, K. Harada, S. Saigo and S. Inoue, "Quantitative Driver Acceptance Modeling for Merging Car at
Highway Junction and Its Application to the Design of Merging Behavior Control," in IEEE Transactions on Intelligent
Transportation Systems, vol. 22, no. 1, pp. 329-340, Jan. 2021, doi: 10.1109/TITS.2019.2957391.
[15] Y. Sun, Z. Fu, C. Sun, Y. Hu and S. Zhang, "Deep Multimodal Fusion Network for Semantic Segmentation Using Remote
Sensing Image and LiDAR Data," in IEEE Transactions on Geoscience and Remote Sensing, vol. 60, pp. 1-18, 2022, Art
no. 5404418, doi: 10.1109/TGRS.2021.3108352.
[16] J. A. del Peral-Rosado, G. Seco-Granados, S. Kim and J. A. López-Salcedo, "Network Design for Accurate Vehicle
Localization," in IEEE Transactions on Vehicular Technology, vol. 68, no. 5, pp. 4316-4327, May 2019, doi:
10.1109/TVT.2019.2906485.
[17] A. Benloucif, A. -T. Nguyen, C. Sentouh and J. -C. Popieul, "Cooperative Trajectory Planning for Haptic Shared Control
Between Driver and Automation in Highway Driving," in IEEE Transactions on Industrial Electronics, vol. 66, no. 12,
pp. 9846-9857, Dec. 2019, doi: 10.1109/TIE.2019.2893864.
[18] W. Jeon, A. Zemouche and R. Rajamani, "Tracking of Vehicle Motion on Highways and Urban Roads Using a Nonlinear
Observer," in IEEE/ASME Transactions on Mechatronics, vol. 24, no. 2, pp. 644-655, April 2019, doi:
10.1109/TMECH.2019.2892700.
[19] H. Yi, K. Guan, D. He, B. Ai, J. Dou and J. Kim, "Characterization for the Vehicle-to-Infrastructure Channel in Urban and
Highway Scenarios at the Terahertz Band," in IEEE Access, vol. 7, pp. 166984-166996, 2019, doi:
10.1109/ACCESS.2019.2953890.
[20] Y. Lee, D. Y. Lee, S. H. Lee and Y. Kim, "A Comparative Study on Model Predictive Control Design for Highway Car-
Following Scenarios: Space-Domain and Time-Domain Model," in IEEE Access, vol. 9, pp. 162291-162305, 2021, doi:
10.1109/ACCESS.2021.3131681.
[21] X. Song and M. Yuan, "Contention-Based Opportunistic Spectrum Access in One-Way Highway Vehicular
Networks," in IEEE Access, vol. 7, pp. 149245-149254, 2019, doi: 10.1109/ACCESS.2019.2944591.
[22] “Mitigation of Blackspots on Highways by the Application of Safe System Approach”, Dinesh
K.Yadava,Sujesh D.Ghodmare, Naveen Kumar, Materials Today: Proceedings, Elsevier,
https://doi.org/10.1016/j.matpr.2021.11.042, online 19 November 2021
[23] Sujesh D. Ghodmare, Akshay Gulghane, An attempt for sustainability evaluation considering implementation
of rapid transit system, Materials Today: Proceedings,2022,ISSN 2214-
7853, https://doi.org/10.1016/j.matpr.2022.01.262.
(https://www.sciencedirect.com/science/article/pii/S2214785322002942)
[24] “Transportation Planning Using Conventional Four Stage Modeling”, Sujesh D. Ghodmare, Geetanjali Yadav
(2021), Turkish Journal of Computer and Mathematics EducationVol.12 No.12 (2021), 2891-2897, Published
online: 23 May 2021
[25] “Application of the multi attribute utility technique with its for sustainability evaluation of emerging
metropolitan city of Nagpur”, International Journal of Civil Engineering and Technology (2019), EID: 2-
s2.0-85060924119, Sujesh D. Ghodmare, Preeti Bajaj, B. V. Khode
[26] Feasibility Study on Widening of Roads as Per Traffic Survey”, Manisha P. Gupta, S. D. Ghodmare, Dr. B. V.
Khode. (2021). Design Engineering, 2021(04), 210 - 217. Vol 2021: Issue 04