ADVANCED RIDGE REGRESSION USING IMFT MODEL FOR RSU DESIGN IN VEHICULAR NETWORKS

Jou nal o Theo e ical and Applied In o ma ion Technology 31s May 2025. Vol.103. No.10 © Li le Lion Scien i ic ISSN: 1992-8645 www.ja i .o g E-ISSN: 1817-3195 4163 ADVANCED RIDGE REGRESSION USING IMFT MODEL FOR RSU DESIGN IN VEHICULAR NETWORKS KRISHNA KOMARAM1, NAGARJUNA KARYEMSETTY2 1Resea ch Schola , Depa men o Compu e Science and Enginee ing, Kone u Lakshmaiah Educa ion Founda ion, Vaddeswa am, Vijayawada, Andh a P adesh, India. Email:k [email protected] 2 Depa men o Compu e Science and Enginee ing, Kone u Lakshmaiah Educa ion Founda ion, Vaddeswa am, Vijayawada, Andh a P adesh, India. Email: naga junak@kluni e si y.in ABSTRACT In he ealm o Vehicula Ad Hoc Ne wo ks (VANETs), Roadside Uni s (RSUs) play a pi o al ole in enhancing communica ion, da a p ocessing, and p edic i e analy ics. This pape in oduces a no el hyb id design ha in eg a es Ridge Reg ession and XG-Boos algo i hms o op imize he da a p ocessing and p edic ion capabili ies o RSUs, aimed a imp o ing a ic managemen and sa e y applica ions. The hyb id amewo k wi h IMFT (in e -in a mod il e ) algo i hm employs Ridge Reg ession o obus ini ial da a p ocessing, minimizing o e i ing and ensu ing eliabili y in he noisy, dynamic en i onmen o ehicula da a. Fea u e ex ac ion wi h IMFT is u ilized o encompass he ele an ea u es be o e u ilizing Ridge- Model o high-accu acy p edic ions, le e aging i s g adien boos ing capabili ies o acili a e imely in e en ions and op imize a ic low. Fu he mo e, he a chi ec u e o he RSU is expanded o include essen ial uni s such as communica ion modules, da a s o age, and use in e ace componen s, all unc ioning cohesi ely o c ea e a comp ehensi e sys em. Wi h he p oposed IMFT design we ha e inco po a ed ex ensi e simula ions wi h K- old loss o demons a e ha he p oposed IMFT wi h Ridge Model design signi ican ly enhances p edic ion accu acy and p ocessing e iciency compa ed o adi ional me hods (Elas ic Ne .) wi h mo e han 98% o imp o ed R2-sco e. By op imizing he ope a ional capabili ies o RSUs in VANETs, his wo k con ibu es o he de elopmen o sma e and sa e u ban mobili y solu ions, pa ing he way o mo e e ec i e a ic managemen and imp o ed ehicula sa e y. Keywo ds: Vehicula Ad Hoc Ne wo ks (VANETs), Roadside Uni s (RSUs), Secu i y P o ocols, Ene gy Managemen , ML (Machine Lea ning), Ridge Reg ession. 1. INTRODUCTION Vehicula Ad Hoc Ne wo ks (VANETs) ep esen a ans o ma i e app oach o imp o ing oad sa e y and a ic managemen by acili a ing communica ion among ehicles and in as uc u e. A he hea o his sys em a e Roadside Uni s (RSUs), which se e as c i ical nodes ha enable da a exchange be ween ehicles, a ic managemen sys ems, and a ious se ices. As u ban en i onmen s g ow inc easingly complex, he demand o e icien communica ion and eal- ime da a p ocessing wi hin VANETs in ensi ies. RSUs a e uniquely posi ioned o b idge he gap be ween ehicles and cen al a ic managemen , o e ing oppo uni ies o enhanced oad sa e y, educed conges ion, and op imized a ic low. Designing e ec i e RSU sys ems in ol es he in eg a ion o ad anced communica ion echnologies and in elligen algo i hms ha can p ocess la ge olumes o da a in eal ime. T adi ional app oaches o en all sho in handling he dynamic and he e ogeneous na u e o ehicula da a. This necessi a es inno a i e me hods ha le e age machine lea ning algo i hms, enabling RSUs o pe o m p edic i e analy ics and decision-making wi h high accu acy. The inco po a ion o such algo i hms can signi ican ly imp o e he esponsi eness o a ic managemen sys ems, ensu ing ha in e en ions a e imely and con ex ually ele an . Jou nal o Theo e ical and Applied In o ma ion Technology 31s May 2025. Vol.103. No.10 © Li le Lion Scien i ic ISSN: 1992-8645 www.ja i .o g E-ISSN: 1817-3195 4164 Cu en ly in RSU’s he use o machine lea ning algo i hms has been p o en mo e e ec i e and in ui i e models o imp o e a ic managemen and sa e y. In hese app oaches he need o ensemble and i s c i e ion’s a e mo e e ec i e and u ilized wi h cu en algo i hm’s such as Random Fo es (hype - uned), is a obus ensemble me hod ha combines mul iple decision ees, making i g ea o handling noisy, complex a ic da a. By uning pa ame e s like he numbe o ees and dep h, i helps RSUs p edic a ic pa e ns, de ec anomalies, and manage conges ion. Ridge Reg ession (hype - uned), on he o he hand, uses L2 egula iza ion o p e en o e i ing and is use ul o p edic ing con inuous a iables like ehicle speed o a ic low. I is e ec i e in handling co ela ed ea u es and ensu ing accu a e, s able p edic ions, hough i assumes linea ela ionships. Simila ly, Linea Reg ession (hype - uned) is simple and in e p e able, helping RSUs o ecas a ic densi y o ehicle speed, bu i may s uggle wi h complex, nonlinea da a. Lasso Reg ession (hype - uned) ocuses on selec ing he mos ele an ea u es, which is c ucial o e icien eal- ime decision-making in VANETs, al hough i can be uns able wi h highly co ela ed ea u es. Elas ic Ne (hype - uned) combines he s eng hs o bo h Lasso and Ridge, making i ideal o si ua ions wi h many co ela ed ea u es, bu i equi es ca e ul uning o wo pa ame e s and can be compu a ionally expensi e. Each algo i hm b ings unique s eng hs o RSU designs in VANETs. Random Fo es is e ec i e in noisy, unp edic able en i onmen s, while Ridge and Lasso a e g ea o ea u e selec ion and a oiding o e i ing. Howe e , hese algo i hms ha e hei limi a ions, such as assump ions o linea i y o compu a ional in ensi y. The hyb id app oach, combining Ridge Reg ession o p e- p ocessing and XGBoos o p edic ion, o e comes hese gaps by i s il e ing ou noise and i ele an ea u es be o e applying XG-Boos 's powe ul p edic ion capabili ies. This combina ion p o ides mo e accu a e, e icien a ic managemen , pa icula ly in dynamic and complex en i onmen s like VANETs, whe e eal- ime p edic ions and decision-making a e c i ical o imp o ing oad sa e y and op imizing a ic low. 1.1 P oblem S a emen In he con ex o Vehicula Ad Hoc Ne wo ks (VANETs), Roadside Uni s (RSUs) a e c ucial o acili a ing communica ion be ween ehicles and a ic managemen sys ems o enhance oad sa e y and op imize a ic low. Howe e , he dynamic, noisy, and complex na u e o a ic da a poses signi ican challenges o RSUs in making accu a e, eal- ime p edic ions o a ic managemen and sa e y in e en ions. T adi ional app oaches o p ocessing and p edic ing a ic pa e ns o en all sho due o hei inabili y o handle la ge olumes o da a, complex in e ac ions, and he he e ogeneous na u e o he in o ma ion. Fo example, basic eg ession models may no e ec i ely cap u e nonlinea ela ionships be ween a ious a ic ac o s, and models like Random Fo es may become compu a ionally expensi e when dealing wi h la ge da ase s. Addi ionally, issues like o e i ing, da a spa si y, and he need o eal- ime decision-making u he complica e he de elopmen o e icien RSU sys ems. To add ess hese challenges, he p oposed solu ion is a hyb id model ha in eg a es Ridge Reg ession o obus da a p e-p ocessing and XGBoos o accu a e p edic ions. The Ridge Reg ession componen helps in educing he complexi y o he model and p e en ing o e i ing, especially in noisy en i onmen s, while XGBoos 's g adien boos ing capabili ies a e le e aged o high- accu acy a ic p edic ions. This hyb id app oach no only enhances he abili y o RSUs o p ocess and p edic a ic condi ions bu also ensu es imely and accu a e in e en ions, con ibu ing o sa e and mo e e icien a ic managemen . The hyb id model o e comes he limi a ions o adi ional machine lea ning algo i hms by combining he s eng hs o Ridge Reg ession and XGBoos , add essing issues such as noise, ea u e selec ion, and he need o eal- ime decision- making in he dynamic con ex o VANETs. 1.2 Challenges Implemen ing e ec i e RSU sys ems wi hin VANETs p esen s se e al key challenges. Fi s , ensu ing da a p i acy and secu i y is pa amoun , as sensi i e in o ma ion is exchanged be ween ehicles and in as uc u e. Second, communica ion la ency can dis up he imely exchange o c i ical da a, unde mining he esponsi eness o a ic managemen in e en ions. Las ly, he in eg a ion o di e se da a Jou nal o Theo e ical and Applied In o ma ion Technology 31s May 2025. Vol.103. No.10 © Li le Lion Scien i ic ISSN: 1992-8645 www.ja i .o g E-ISSN: 1817-3195 4165 sou ces om a ious ehicles necessi a es sophis ica ed algo i hms capable o p ocessing his in o ma ion in eal ime, which can be a signi ican echnical hu dle. E en hough hese challenges e ec he design cons ain s he p oposed model end o p o ide he esul ing solu ion wi h in eg a ed app oach wi h wo powe ul machine lea ning algo i hms: Ridge Reg ession and XGBoos o op imize he da a p ocessing and p edic ion capabili ies o Roadside Uni s (RSUs) in Vehicula Ad Hoc Ne wo ks (VANETs). The eason o using hese algo i hms is o add ess he complex and dynamic na u e o ehicula da a, which can be noisy and he e ogeneous. 1.3 P oposed Concep The IMFT (In e -In a Model Fil e ing Technique) model, in eg a ed wi h Ridge Reg ession, in oduces a no el hyb id machine lea ning app oach o op imizing Roadside Uni (RSU) pe o mance in Vehicula Ad-hoc Ne wo ks (VANETs). Unlike con en ional RSU deploymen models ha ely on heu is ic o s a ic op imiza ion echniques, he IMFT model dynamically il e s and p ocesses a ic, ene gy, and memo y da a using a wo- ie il e ing mechanism. Ini ially, he model pa i ions da a in o in a- ehicle (D_in a) and ex a- ehicle (D_ex a) da ase s, ensu ing ha localized ehicle pa e ns a e analyzed sepa a ely om b oade a ic in luences. Ridge Reg ession is hen applied o minimize noise, p e en o e i ing, and imp o e gene aliza ion, allowing he model o accu a ely classi y and o ecas c i ical RSU pa ame e s such as bandwid h alloca ion, powe consump ion, and memo y u iliza ion. The no el y o his app oach lies in i s dual-laye e o il e ing, whe e p edic ions a e alida ed agains ex a- ehicle da a, and only hose wi hin an accep able h eshold (ϵ) a e e ained o u he analysis. This p ocess signi ican ly enhances da a eliabili y, educing he likelihood o e oneous p edic ions, which is a common issue in pu ely AI-d i en RSU op imiza ion models. Addi ionally, by inco po a ing ensemble lea ning echniques such as s acking and boos ing, he IMFT model e ines he accu acy o Ridge Reg ession ou pu s by combining hem wi h decision ees and g adien boos ing models. This ensu es a holis ic, da a- d i en RSU design ha no only op imizes eal- ime communica ion and esou ce alloca ion bu also adap s dynamically o a ic a ia ions, making i mo e obus compa ed o adi ional machine lea ning-based RSU managemen sys ems. The p ima y aim o his s udy is o de elop an in elligen , scalable, and e icien RSU op imiza ion amewo k ha le e ages Ridge Reg ession and IMFT il e ing o imp o e da a ansmission speed, ba e y powe u iliza ion, and memo y managemen in dynamic ehicula en i onmen s. Unlike p e ious s udies ha ocus on isola ed machine lea ning models, he p oposed hyb id IMFT app oach enhances o ecas ing accu acy by il e ing ou high-e o p edic ions, ensu ing adap i e and eal- ime RSU decision-making. Ou come measu es include inc eased ansmission e iciency, wi h RSUs dynamically adjus ing bandwid h alloca ion based on o ecas ed conges ion, leading o lowe la ency and imp o ed V2I communica ion eliabili y. Fu he mo e, he model ex ends RSU ba e y li e by p edic ing ene gy demand and op imizing powe -sa ing modes, pa icula ly in enewable-ene gy-powe ed RSUs. In e ms o memo y managemen , he IMFT model enhances caching e iciency h ough ein o cemen lea ning- d i en memo y p io i iza ion, ensu ing ha only he mos ele an ehicula da a is s o ed, p e en ing o e load. These ad ancemen s es ablish he no el con ibu ion o he s udy, demons a ing how an e o - il e ed Ridge Reg ession model, combined wi h ensemble lea ning, can signi ican ly enhance RSU pe o mance. By in eg a ing eal- ime o ecas ing wi h adap i e decision-making, he p oposed sys em su passes con en ional me hods, p o iding a scalable, AI-d i en solu ion o sma ehicula in as uc u e ha ensu es e icien , eal- ime a ic managemen and op imized esou ce u iliza ion in VANETs. 1.4 Objec i es 1. To de elop a hyb id design o RSUs ha in eg a es machine lea ning algo i hms o enhanced da a p ocessing and p edic i e capabili ies. 2. To e alua e he pe o mance o P oposed- IMFT Ridge Reg ession in eal- ime a ic analysis and decision-making wi hin VANETs. 3. To add ess da a p i acy, communica ion la ency, and in eg a ion challenges o c ea e a obus RSU amewo k ha suppo s sma e u ban mobili y solu ions. Jou nal o Theo e ical and Applied In o ma ion Technology 31s May 2025. Vol.103. No.10 © Li le Lion Scien i ic ISSN: 1992-8645 www.ja i .o g E-ISSN: 1817-3195 4166 Ou line o he Pape : This pape p esen s a hyb id model design aimed a enhancing Roadside Uni s (RSUs) in Vehicula Ad Hoc Ne wo ks (VANETs) h ough a sequence o well-de ined sec ions. Sec ion 1 - In oduc ion in oduces he concep o VANETs and he pi o al ole o RSUs in a ic managemen and sa e y, ou lining challenges such as handling dynamic and noisy da a, along wi h he objec i e o imp o ing eal- ime p edic ion accu acy. In Sec ion 2 - Li e a u e Su ey, a e iew o exis ing me hods is p o ided, ocusing on algo i hms like Random Fo es , Ridge Reg ession, and Linea Reg ession, iden i ying gaps such as o e i ing, high compu a ional cos s, and he inabili y o model nonlinea i ies. Sec ion 3 - Exis ing Me hods A. Algo i hms di es deepe in o hese algo i hms, explaining hei s eng hs and weaknesses in a ic p edic ion asks, which mo i a es he need o a mo e obus app oach. Sec ion 4 - Me hodology p esen s he p oposed hyb id app oach, combining Ridge Reg ession o da a p ep ocessing wi h XGBoos o high-accu acy p edic ions. This sec ion elabo a es on he bene i s o in eg a ing hese algo i hms o o e come he limi a ions o exis ing me hods, ensu ing be e pe o mance in eal- ime, noisy a ic en i onmen s. Sec ion 5 - Resul s and Discussion p esen s he ou comes o simula ions, compa ing he p oposed model wi h exis ing me hods and demons a ing signi ican imp o emen s in p edic ion accu acy and ope a ional e iciency. Finally, he Conclusion and Scope sec ion summa izes he e ec i eness o he p oposed model and ou lines u u e oppo uni ies o enhance RSU sys ems o sma e a ic managemen and imp o ed oad sa e y. 2. LITERATURE SURVEY In ecen yea s, he deploymen and op imiza ion o Roadside Uni s (RSUs) in Vehicula Ad Hoc Ne wo ks (VANETs) ha e gained signi ican a en ion due o hei po en ial o enhance communica ion, sa e y, and a ic managemen . Yu e al. (2022) [1] p opose an RSU deploymen s a egy ha aligns wi h a ic demand, highligh ing he impo ance o op imizing in as uc u e in es men s while ensu ing oad sa e y. Thei indings emphasize a da a-d i en app oach o RSU placemen , which can e ec i ely mi iga e delays and enhance o e all ne wo k e iciency. Gao e al. (2021) [2] explo e he op imal RSU deploymen p oblem using app oxima ion algo i hms. They p esen wo g eedy algo i hms ha yield igh app oxima ion a ios, showcasing he e ec i eness o heu is ic me hods in add essing deploymen challenges in one-dimensional VANETs. Thei con ibu ions lie in demons a ing how g eedy app oaches can simpli y complex op imiza ion p oblems, p o iding a p ac ical amewo k o e icien RSU deploymen . Yang e al. (2020) [3] in oduce an analy ical model o ene gy-ha es ing RSUs wi h a dynamic se ice adius, add essing he c i ical aspec o ene gy managemen in VANETs. By modeling ehicle dynamics and ba e y pe o mance, hei indings illus a e how s a egically deploying ene gy- ha es ing RSUs can ex end se ice a eas and imp o e connec i i y. This wo k unde sco es he po en ial o in eg a ing enewable ene gy solu ions in u u e ehicula ne wo ks. Yada e al. (2024) [4] ocus on secu i y wi hin VANETs by p esen ing an imp o ed ehicle- o- og au hen ica ion sys em. Thei s udy emphasizes he need o secu e communica ion p o ocols in oad condi ion moni o ing, u ilizing a obus au hen ica ion mechanism ha balances e iciency and secu i y. The p oposed sys em con ibu es o enhancing us among ehicles and in as uc u e, which is c ucial o he success ul implemen a ion o in elligen anspo a ion sys ems. Ud Din e al. (2021) [5] p opose a caching s a egy o ehicula ne wo ks based on In o ma ion-Cen ic Ne wo king (ICN), highligh ing he ole o caching in enhancing da a a ailabili y and educing la ency. Thei indings indica e ha an e icien caching mechanism can signi ican ly imp o e da a e ie al imes, con ibu ing o sa e and mo e esponsi e ehicula communica ions. Sepasgoza and Pie e (2022) [6] in oduce a ne wo k a ic p edic ion model ha u ilizes a i icial in elligence me hods o accoun o a ious oad a ic pa ame e s. Thei app oach in eg a es machine lea ning algo i hms o enhance he accu acy o a ic p edic ions, he eby acili a ing be e a ic managemen and ou ing decisions. This s udy con ibu es o he g owing body o li e a u e ha emphasizes he ole o p edic i e analy ics in op imizing VANET ope a ions. Zhao e al. (2022) [7] p esen a collabo a i e da a co ec ion me hod o Vehicle- o-E e y hing Jou nal o Theo e ical and Applied In o ma ion Technology 31s May 2025. Vol.103. No.10 © Li le Lion Scien i ic ISSN: 1992-8645 www.ja i .o g E-ISSN: 1817-3195 4167 (V2X) communica ions, ocusing on imp o ing oad sa e y h ough accu a e da a sha ing. Thei indings e eal ha collabo a i e da a co ec ion can signi ican ly enhance he eliabili y o sa e y messages, he eby educing he likelihood o acciden s caused by inco ec o ou da ed in o ma ion. And eou e al. (2023) [8] explo e he use o UAV-assis ed RSUs o V2X connec i i y, le e aging Vo onoi diag ams o op imize he co e age o unmanned ae ial ehicles in 6G+ in as uc u es. Thei esea ch con ibu es o he in eg a ion o ae ial ne wo ks wi h e es ial ones, sugges ing ha UAVs can e ec i ely complemen adi ional RSU deploymen s, pa icula ly in challenging u ban en i onmen s. Bashi e al. (2024) [9] de elop a no el Cu e C ash A oidance P o ocol (2CAP) aimed a minimizing acciden s in VANETs. Thei wo k unde sco es he impo ance o in elligen sensing and communica ion in eal- ime acciden p e en ion, con ibu ing o he design o sa e ehicula en i onmen s h ough ad anced p o ocol de elopmen . Thumbu e al. (2021) [10] p esen a ce i ica eless agg ega e signa u e-based au hen ica ion scheme ha enhances secu i y and e iciency in ehicula ne wo ks. Thei con ibu ions ocus on educing compu a ional o e head while ensu ing obus au hen ica ion, add essing a c i ical ba ie o widesp ead adop ion o VANET echnologies. Ali (2023) [11] discusses a og-based g een VANET in as uc u e ha emphasizes ene gy e iciency and sus ainable p ac ices. By in eg a ing og compu ing wi h RSU design, he s udy highligh s he po en ial o educing he ca bon oo p in o ehicula ne wo ks, pa ing he way o mo e en i onmen ally iendly solu ions in in elligen anspo a ion sys ems. Xia e al. (2023) [12] u ilize ein o cemen lea ning o de elop an in o ma ion dissemina ion policy in VANETs, demons a ing he applicabili y o ad anced machine lea ning echniques in op imizing da a ansmission s a egies. Thei indings con ibu e o a be e unde s anding o adap i e communica ion p o ocols in dynamic ehicula en i onmen s. Chen e al. (2022) [13] in oduce a mul isigna u e-based eme gency epo ing scheme, ocusing on enhancing secu i y and e iciency in acciden scena ios. Thei wo k emphasizes he necessi y o eliable and quick communica ion du ing eme gencies, he eby con ibu ing o he o e all sa e y mechanisms in VANETs. Aman e al. (2021) [14] p esen a p i acy- p ese ing au hen ica ion p o ocol ailo ed o he In e ne o Vehicles, add essing he c i ical issue o use p i acy in ehicula communica ions. Thei con ibu ions emphasize he impo ance o balancing secu i y and p i acy, p o iding a amewo k o u u e p o ocols ha p io i ize bo h. Wang e al. (2022) [15] de elop an imp o ed ce i ica eless condi ional p i acy-p ese ing au hen ica ion scheme wi h e oca ion capabili ies. This wo k con ibu es o he g owing need o lexible and secu e au hen ica ion me hods in VANETs, ensu ing ha ehicles can communica e sa ely while main aining use anonymi y. Ismail e al. (2024) [16] in es iga e esou ce managemen s a egies in UAV-assis ed VANETs, ocusing on line-o -sigh communica ions o enhance h oughpu and educe in e e ence. Thei indings unde sco e he impo ance o coo dina ed esou ce alloca ion in imp o ing o e all ne wo k pe o mance. Zhou e al. (2024) [17] p opose a cloud-assis ed au hen ica ion key ag eemen p o ocol o VANETs, add essing he challenges o secu e communica ion in highly mobile en i onmen s. Thei s udy highligh s he po en ial o cloud compu ing o enhance au hen ica ion p ocesses, con ibu ing o mo e eliable ehicula communica ions. Liu e al. (2024) [18] in oduce an anonymous aceable and e ocable c eden ial sys em using blockchain echnology o VANETs. Thei indings illus a e how blockchain can p o ide a decen alized app oach o au hen ica ion, add essing p i acy conce ns while enhancing us in ehicle communica ions. Tangade e al. (2020) [19] explo e us managemen schemes based on hyb id c yp og aphy o secu e communica ions in VANETs. Thei con ibu ions highligh he necessi y o es ablishing us among ehicles and in as uc u e, add essing ulne abili ies ha can be exploi ed in open communica ion en i onmen s. Wang and Liu (2021) [20] p esen a secu e and e icien message au hen ica ion p o ocol o VANETs, ocusing on educing he isks o message o ge y and ensu ing da a in eg i y. Thei indings emphasize he impo ance o eliable au hen ica ion me hods in main aining he secu i y o ehicula communica ions. Yan e al. (2024) [21] in oduce an edge-assis ed hie a chical ba ch au hen ica ion scheme o VANETs, showcasing Jou nal o Theo e ical and Applied In o ma ion Technology 31s May 2025. Vol.103. No.10 © Li le Lion Scien i ic ISSN: 1992-8645 www.ja i .o g E-ISSN: 1817-3195 4168 how edge compu ing can enhance he e iciency o au hen ica ion p ocesses. Thei esea ch con ibu es o he de elopmen o scalable and e ec i e secu i y solu ions in ehicula ne wo ks. Al-Sha eeda e al. (2020) [22] discuss e icien condi ional p i acy p ese a ion wi h mu ual au hen ica ion, add essing he dual challenges o secu i y and use p i acy in VANETs. Thei s udy highligh s he need o p o ocols ha can p o ide obus au hen ica ion wi hou comp omising use da a. Huang and Lai (2020) [23] examine delay- cons ained da a o loading in VANETs, u ilizing mul i-access edge compu ing o imp o e ne wo k e iciency. Thei indings e eal he po en ial o edge compu ing a chi ec u es o acili a e imely da a sha ing, he eby enhancing he pe o mance o ehicula applica ions. Baza e al. (2022) [24] ocus on de ec ing Sybil a acks in VANETs using p oo s o wo k and loca ion. Thei esea ch unde sco es he impo ance o eliable iden i y e i ica ion mechanisms in p e en ing malicious ac i i ies ha can unde mine he in eg i y o ehicula ne wo ks. Finally, Zhang e al. (2022) [25] p opose a us - based and p i acy-p ese ing pla oon ecommenda ion scheme, highligh ing he ole o us in acili a ing coope a i e manoeu es among ehicles. Thei wo k emphasizes he impo ance o bo h p i acy and secu i y in de eloping e ec i e pla ooning s a egies in VANETs, con ibu ing o sa e and mo e e icien anspo a ion sys ems. Collec i ely, hese s udies o e a comp ehensi e o e iew o he ad ancemen s in RSU deploymen , secu i y p o ocols, and communica ion s a egies in VANETs, pa ing he way o u u e esea ch and implemen a ion in in elligen anspo a ion sys ems. Summa y: The p oposed hyb id RSU op imiza ion model, in eg a ing Ridge Reg ession o obus da a p ep ocessing and XGBoos o p edic i e analy ics, signi ican ly imp o es upon exis ing VANET esea ch. Unlike p io s udies such as Sepasgoza and Pie e (2022) [6], which ely solely on AI-d i en a ic p edic ions, his app oach i s employs Ridge Reg ession o il e ou noisy da a and p e en o e i ing. This ensu es ha only ele an a ic ea u es a e conside ed be o e passing he da a o XGBoos , which i e a i ely e ines p edic ions o a ic low and haza d de ec ion. Compa ed o Gao e al. (2021) [2], who u ilized heu is ic me hods o RSU placemen , his model o e s a mo e adap i e, da a-d i en app oach ha can dynamically adjus o eal- ime a ic condi ions. Addi ionally, while s udies like Yu e al. (2022) [1] op imize RSU deploymen based on a ic demand, hey lack an in eg a ed p edic i e laye . The p oposed hyb id model no only places RSUs op imally bu also enhances hei unc ionali y by p edic ing conges ion, enabling p oac i e in e en ions such as adap i e signal imings and e ou ing. Fu he mo e, compa ed o Xia e al. (2023) [12], who applied ein o cemen lea ning o in o ma ion dissemina ion, his app oach le e ages a hyb id ML model ha enhances bo h ea u e selec ion and p edic ion accu acy, making i mo e e icien in handling dynamic ehicula en i onmen s.Despi e i s ad an ages, he model also p esen s challenges, p ima ily in e ms o compu a ional complexi y and da a p i acy. Unlike Yang e al. (2020) [3], who ocus on ene gy-ha es ing RSUs, his app oach demands signi ican p ocessing powe , making implemen a ion di icul in esou ce-cons ained en i onmen s. Addi ionally, handling la ge-scale ehicula da a aises p i acy conce ns, as highligh ed by Yada e al. (2024) [4], equi ing obus enc yp ion and anonymiza ion echniques. Ano he limi a ion is po en ial la ency, as eal- ime XGBoos compu a ions may in oduce delays, especially in high-speed ehicula scena ios, whe eas Chen e al. (2022) [13] emphasize he necessi y o low-la ency eme gency communica ion. To mi iga e hese challenges, u u e esea ch could in eg a e edge compu ing o decen alized p ocessing, educing compu a ional load while main aining p edic i e accu acy. Mo eo e , sel -lea ning RSUs ha dynamically adjus con igu a ions based on e ol ing a ic pa e ns could u he enhance scalabili y. In con as o s a ic RSU placemen models, his app oach enables p oac i e, da a-d i en a ic managemen , making i a supe io choice o sma u ban mobili y solu ions, p o ided ha in as uc u e and secu i y conce ns a e adequa ely add essed. 3. EXISTNG METHOD The VANET sys em has wo communica ion modes: ehicle- o- ehicle (V2V) and ehicle- o- Jou nal o Theo e ical and Applied In o ma ion Technology 31s May 2025. Vol.103. No.10 © Li le Lion Scien i ic ISSN: 1992-8645 www.ja i .o g E-ISSN: 1817-3195 4169 in as uc u e (V2I) communica ion. Thei communica ion echnology is dedica ed sho - ange communica ion (DSRC), as de ined in IEEE 802.11p and IEEE 1609.4 DSRC wi eless access in ehicle con ex s (WAVE) [33, 34]. Vehicles go h ough he oad ne wo k wi h a speci ic des ina ion. We assume ha each ehicle will send in o ma ion o he des ina ion, such as eques ing pa king in o ma ion and subsc ibing o s o e in o ma ion. We conside da a deli e y h ough he G eedy Pe ime e S a eless Rou ing (GPSR) P o ocol [4] because he low o in o ma ion can always be expec ed o ind he sho es pa h o ansmission based on he in ol ed loca ions. The da a deli e y p ocedu e and ehicle mo emen a e depic ed in Figu e 1. Vehicle V1 will anspo da a o Pa king L o F. The ideal da a ansmission pa h in VANETs is V1→ V2→V3→V4→F. In gene al, da a is ansmi ed om V1 o F. The pa h can be ep esen ed by he in e sec ing sequence, which is simpli ied o Ia→ Ib→ Ic. We can sepa a e his pa h. Da a om an Ia o Ib can pass h ough ehicles (V1→V2→V3).Howe e , when he dis ance be ween V3 and V4 is oo g ea and exceeds he communica ion ange, da a canno be ans e ed om Ib o Ic unless V3 accele a es nea V4 o mo es nea F. I he RSU is loca ed in he oad sec ion, he da a deli e y pa h will be V3→RSU→V4→F as i is wi hin communica ion ange o V3 and V4. Fig.1: Rep esen a ion o he Ensemble XGBOOST and RFC model based VANET RSU design Wi eless da a ans e o e an RSU is subs an ially mo e e icien han da a packe deli e y ia a ehicle [3]. Ob iously, using RSUs as a elay o con ey da a is mo e e icien han doing so h ough ehicle mo emen . 3.1 Algo i hms 3.1.1 Random Fo es (Hype -Tuned) - RSU in VANET Random Fo es is an ensemble lea ning me hod ha builds mul iple decision ees and combines hei p edic ions. In VANET, his model can be used o asks such as classi ying ehicle Beha iou s, de ec ing anomalies in a ic pa e ns, o p edic ing a ic conges ion. Hype pa ame e uning in Random Fo es ocuses on op imizing pa ame e s like he numbe o ees (n_es ima o s), he maximum dep h o ees (max_dep h), and he numbe o ea u es conside ed du ing spli ing (max_ ea u es). These uned hype pa ame e s ensu e he model a oids o e i ing while s ill cap u ing he in ica e ela ionships in he da a, such as ehicle mo emen , oad condi ions, and communica ion pa e ns. In RSU design wi hin a VANET, hype - uned Random Fo es can be used o imp o e a ic p edic ion and con ol. The RSU sys em can ga he da a om ehicles in i s icini y (e.g., speed, loca ion, oad condi ions) and use Random Fo es Jou nal o Theo e ical and Applied In o ma ion Technology 31s May 2025. Vol.103. No.10 © Li le Lion Scien i ic ISSN: 1992-8645 www.ja i .o g E-ISSN: 1817-3195 4170 o p edic and manage a ic low o ehicle sa e y. Fo example, i can p edic po en ial acciden s o oad conges ion, enabling he RSU o ake ac ions like e ou ing ehicles o issuing wa nings. The ensemble na u e o Random Fo es makes i obus o noisy da a, which is common in VANETs due o he dynamic and unp edic able na u e o ehicle ne wo ks. 3.1.2 Ridge Reg ession (Hype -Tuned) - RSU in VANET Ridge eg ession is a linea eg ession model ha applies L2 egula iza ion o educe model complexi y and p e en o e i ing. In VANET, Ridge eg ession can be used o con inuous p edic ions, such as es ima ing ehicle speed, a ic low, o uel consump ion based on a a ie y o ea u es (e.g., oad ype, wea he , o ehicle speci ica ions). Hype - uning Ridge eg ession p ima ily ocuses on op imizing he egula iza ion pa ame e (alpha) o ensu e he model gene alizes well o new da a and doesn' become o e ly sensi i e o any one ea u e. In RSU design o VANETs, hype - uned Ridge eg ession can help imp o e p edic ion accu acy o a ic- ela ed pa ame e s, enabling be e a ic managemen . Fo ins ance, i can p edic ehicle speed o expec ed a el ime ac oss di e en oad segmen s. By using op imal hype pa ame e s, Ridge eg ession ensu es he RSU can accu a ely model con inuous da a ends like ehicle Beha iou ac oss a oad ne wo k, helping o manage esou ces e ec i ely and p o ide accu a e in o ma ion o d i e s, which is c ucial o sa e y, e iciency, and ou e op imiza ion. 3.1.3 Linea Reg ession (Hype -Tuned) - RSU in VANET Linea eg ession is a undamen al model ha es ablishes a ela ionship be ween inpu ea u es and a con inuous a ge a iable. In VANET, i can be used o p edic a iables such as a ic densi y, ehicle speed, o signal s eng h in communica ion ne wo ks. Hype - uning Linea Reg ession in ol es adjus ing pa ame e s o implemen ing echniques like Lasso o Ridge egula iza ion o op imize pe o mance and p e en unde i ing o o e i ing, especially when dealing wi h noisy da a. In an RSU sys em o VANET, Linea Reg ession can be used o o ecas ehicle a i al imes a in e sec ions, p edic a ic olume o a gi en a ea, o es ima e ehicle emissions based on a ic pa e ns. Hype - uning his model ensu es ha he p edic ions made by he RSU a e accu a e and eliable, aiding in eal- ime decision-making o managing a ic and imp o ing oad sa e y. Tuning helps a oid poo p edic ions ha could a ise due o o e simpli ica ion o excessi e complexi y, p o iding mo e eliable inpu s o he RSU o ake ac ion on. 3.1.4 Lasso Reg ession (Hype -Tuned) - RSU in VANET Lasso eg ession, which uses L1 egula iza ion, encou ages spa si y in he model by penalizing he coe icien s o less impo an ea u es. In VANET, Lasso eg ession can be use ul o scena ios whe e only a subse o ea u es is ele an o p edic ing a ic condi ions o ehicle beha iou . Hype - uning he Lasso model ocuses on op imizing he egula iza ion pa ame e (alpha) o ind he igh balance be ween i ing he da a well and selec ing only he mos impo an ea u es. In he con ex o RSU in VANET, Lasso eg ession can help iden i y he mos ele an ac o s ha a ec a ic low, ehicle speed, o signal s eng h in he ne wo k. By selec ing he mos impo an ea u es, Lasso enables he RSU o ope a e mo e e icien ly, ensu ing ha a ic managemen decisions a e based on he mos in luen ial a iables. Hype - uned Lasso eg ession helps educe complexi y in he model, which is impo an o eal- ime p ocessing in a VANET, whe e quick and e icien decision-making is equi ed o sa e y and a ic op imiza ion. 3.1.5 Elas ic Ne (Hype -Tuned) - RSU in VANET Elas ic Ne combines he bene i s o bo h L1 (Lasso) and L2 (Ridge) egula iza ion, making i use ul when dealing wi h a la ge numbe o co ela ed ea u es, which is o en he case in VANET da a. Elas ic Ne can be pa icula ly e ec i e o p edic ing complex ou comes, such as ehicle- o- ehicle communica ion quali y, oad sa e y condi ions, o a ic low, whe e mul iple in e ela ed ea u es need o be conside ed. Hype - uning Elas ic Ne in ol es op imizing bo h he egula iza ion s eng h (alpha) and he mix a io (l1_ a io), which con ols he balance be ween Lasso and Ridge penal ies.In RSU sys ems o VANET, hype - uned Elas ic Ne can be used o mo e accu a e p edic ions and analysis, such as o ecas ing conges ion le els o op imizing Jou nal o Theo e ical and Applied In o ma ion Technology 31s May 2025. Vol.103. No.10 © Li le Lion Scien i ic ISSN: 1992-8645 www.ja i .o g E-ISSN: 1817-3195 4171 communica ion p o ocols o ehicle sa e y. Since VANET da a o en con ains co ela ed ea u es (e.g., a ic condi ions, ehicle speeds, en i onmen al ac o s), Elas ic Ne ’s abili y o handle mul icollinea i y makes i an ideal choice. Tuning he model ensu es ha he RSU can make be e decisions ega ding ou e ecommenda ions, a ic managemen , and acciden p e en ion, ul ima ely enhancing he pe o mance and sa e y o he en i e VANET in as uc u e. 3.2 E alua ing Machine Lea ning Model o RSU In eg a ion in VANET In he con ex o a Vehicula Ad-hoc Ne wo k (VANET) sys em, he Road Side Uni (RSU) plays a pi o al ole in enabling seamless communica ion be ween ehicles and he in as uc u e, ensu ing e icien a ic managemen , sa e y, and imely da a deli e y. To achie e his, se e al machine lea ning algo i hms—Random Fo es , Ridge Reg ession, Linea Reg ession, Lasso Reg ession, and Elas ic Ne —a e implemen ed o p edic a ic condi ions, ehicle beha iou , and communica ion quali y. These algo i hms can be u ilized in wo dis inc ways: wi h hype pa ame e uning o op imal model pe o mance and wi hou uning o quicke , less p ecise esul s. The implemen a ion o hese models is cen al o enhancing he decision-making p ocess, such as adjus ing a ic signals, p edic ing a ic conges ion, o managing eal- ime ehicle ou ing. Hype pa ame e uning is c i ical o imp o ing he accu acy and eliabili y o hese algo i hms. Fo ins ance, in a Random Fo es model, hype pa ame e uning ocuses on op imizing he numbe o ees (n_es ima o s), maximum dep h (max_dep h), and he numbe o ea u es (max_ ea u es). In VANET applica ions, his uned model helps he RSU p edic u u e a ic condi ions, de ec anomalies in ehicle beha iou , and o ecas po en ial acciden s by analyzing ehicle da a like speed, loca ion, and oad condi ions. Simila ly, Ridge Reg ession (wi h L2 egula iza ion) helps p edic con inuous alues such as ehicle speeds and a el imes. Th ough hype pa ame e op imiza ion, i ensu es he model gene alizes well o unseen da a, allowing RSUs o make eal- ime decisions o e icien a ic low and sa e y in e en ions. Elas ic Ne , combining he bene i s o bo h Lasso (L1) and Ridge (L2) egula iza ion, is pa icula ly e ec i e in handling co ela ed ea u es in a ic da a, such as ehicle speed, oad condi ions, and en i onmen al ac o s, o e ing enhanced p edic i e accu acy. When hype - uned, Lasso Reg ession and Linea Reg ession also p o ide alue by ocusing on key ac o s in luencing a ic pa e ns and ou e op imiza ion, hus enabling he RSU o manage ehicle lows mo e e ec i ely. Howe e , un uned models p esen ce ain limi a ions. Fo ins ance, Random Fo es wi hou uning migh misclassi y a ic s a es, leading o ine iciencies in ou ing decisions o ailu e o de ec c ucial anomalies in ime. Simila ly, un uned Ridge Reg ession may esul in o e i ing o unde i ing due o a poo ly op imized egula iza ion pa ame e (alpha), a ec ing p edic ions like a ic low o ehicle speeds. In he case o Linea Reg ession, he lack o uning can lead o o e simpli ica ion o sensi i i y o noise, which migh hinde he RSU’s abili y o manage a ic e ec i ely. The same issue can be obse ed in Lasso Reg ession, whe e an un uned model could ei he o e -penalize i ele an ea u es o ail o p io i ize he mos impo an a iables o a ic managemen . Elas ic Ne also su e s wi hou uning, as i may ail o p ope ly balance he L1 and L2 egula iza ion s eng hs, leading o less e ec i e handling o co ela ed da a, which is p e alen in VANET sys ems. Ul ima ely, un uned models, while s ill unc ional, may lead o ine icien a ic managemen and subop imal da a ou ing, limi ing he ull po en ial o VANET sys ems. When analyzing he pe o mance o hese algo i hms, a ious me ics such as Mean Squa ed E o (MSE), Roo Mean Squa ed E o (RMSE), Mean Absolu e E o (MAE), and R² Sco e p o ide insigh s in o hei accu acy. F om he esul s, we obse e ha while he hype - uned models o Random Fo es , Ridge, Linea Reg ession, Lasso, and Elas ic Ne all ha e ela i ely low MSE and RMSE alues (a ound 867-870 o MSE and 29.45-29.51 o RMSE), hei R² Sco e is close o ze o, indica ing ha hese models a e s uggling o explain he a iance in he da a e ec i ely. The MAE alues a e simila ly close o each o he (a ound 25.47-25.55), showing consis en pe o mance ac oss hese models. The low R² sugges s ha , despi e he models’ ine- uning, hey migh no ully cap u e he complex pa e ns in VANET sys ems, indica ing ha Jou nal o Theo e ical and Applied In o ma ion Technology 31s May 2025. Vol.103. No.10 © Li le Lion Scien i ic ISSN: 1992-8645 www.ja i .o g E-ISSN: 1817-3195 4178 𝑦= ∑𝛼𝑦   Eq: (11) Whe e:  𝑦a e he p edic ions om indi idual models (Ridge, SVM)  Whe e 𝛼 a e he weigh s assigned based on he espec i e model's pe o mance me ics u ilized in p oposed design. Fo p edic ing da a ansmission speedEq.(6), Ridge eg ession is used o model he ela ionship be ween inpu ea u es such as ehicle densi y, a ic load, and his o ical ansmission speeds. The ma ix o inpu ea u es is mul iplied by a coe icien ec o (β), and egula iza ion is in oduced h ough he pa ame e (λ) o p e en o e i ing by penalizing la ge coe icien alues. The L2 no m o he coe icien ec o (‖β‖₂) ensu es he model gene alizes well o new da a. To imp o e p edic ion accu acy, an ensemble me hod is used whe e p edic ions om he Ridge classi ie a e combined wi h o he classi ie s like Decision T ees, as shown in Eq.(7). He e, he model p edic ions om each classi ie a e weigh ed acco ding o hei pe o mance, wi h weigh s (w₁, w₂, ..., wₖ) being assigned o each indi idual model based on hei p edic i e accu acy o impo ance. This weigh ing helps op imize he o e all model by gi ing highe impo ance o be e -pe o ming models, hus enhancing he eliabili y o he p edic ions. Simila ly, o ba e y powe u iliza ion p edic ion Eq.(8), Ridge eg ession models he ela ionship be ween inpu ea u es such as a ic load, en i onmen al condi ions, and RSU con igu a ion. The egula iza ion e m (λ) con ols o e i ing by penalizing la ge coe icien alues, wi h he L2 no m (‖β‖₂) ensu ing ha he model emains obus and gene alizes well o new da a. In he hyb id amewo k o ba e y p edic ion, Eq.(9) shows how he p edic ions om mul iple classi ie s (Random Fo es and G adien Boos ing) a e a e aged, whe e he weigh s assigned o each classi ie e lec i s p edic i e pe o mance. Las ly, o memo y space occupa ion (Eq.(10)), Ridge eg ession p edic s he amoun o memo y u ilized based on ea u es like da a p ocessing load and caching s a egies. The egula iza ion pa ame e (λ) and L2 no m (‖β‖₂) ensu e ha he model a oids o e i ing while cap u ing he ele an ela ionships. In he hyb id app oach o memo y p edic ion, as shown in Eq.(11), p edic ions om indi idual models like Ridge and Suppo Vec o Machines (SVM) a e combined, wi h weigh s assigned based on each model's pe o mance. These weigh s allow he model o adjus he in luence o each classi ie , ensu ing he hyb id design p o ides mo e accu a e and eliable p edic ions o memo y u iliza ion. The ca e ul weigh ing o hese classi ie s based on pe o mance me ics ensu es ha he hyb id model le e ages he s eng hs o each algo i hm, esul ing in mo e obus and p ecise p edic ions o key pe o mance me ics in RSUs. 4.6 IMPLEMENTATION 4.6.1 Expe imen al Se up The expe imen al esul s u ilizing he da ase gene a ed om he speci ied ehicle a ibu es can be explained h ough he alues and hei implica ions o da a ansmission speed, ba e y powe u iliza ion, and memo y space occupa ion in RSUs. 4.6.1.1 Da a T ansmission Speed In he con ex o da a ansmission speed, key ea u es om he ehicle a ibu es, such as speed, loca ion, and b ake s a us, we e pi o al. Fo ins ance, ehicles a eling a highe speeds (up o 120 km/h) gene a e mo e da a due o hei apid changes in loca ion and s a us. The RSUs can use his in o ma ion o p io i ize communica ion o hese ehicles, op imizing bandwid h alloca ion dynamically. The loca ion alues (la i ude and longi ude) help de e mine he densi y o ehicles in speci ic a eas. I a high concen a ion o ehicles is de ec ed in a egion, he RSU can p e-emp i ely alloca e mo e bandwid h o minimize la ency du ing peak a ic condi ions. Addi ionally, he b ake s a us o ehicles can indica e po en ial sa e y e en s, p omp ing immedia e da a ansmission o ale s and communica ion wi h o he ehicles. The in eg a ion o hese ea u es in o he ensemble lea ning model signi ican ly enhances he RSU's abili y o o ecas and adap o eal- ime a ic condi ions, ensu ing as e and mo e eliable communica ion. 4.6.1.2 Ba e y Powe U iliza ion Ba e y powe u iliza ion is in luenced by ea u es like ba e y ol age, engine s a us, and uel le el. Fo example, a lowe ba e y ol age (below 12.0 ol s) could indica e ha he RSU needs o swi ch Jou nal o Theo e ical and Applied In o ma ion Technology 31s May 2025. Vol.103. No.10 © Li le Lion Scien i ic ISSN: 1992-8645 www.ja i .o g E-ISSN: 1817-3195 4179 o an ene gy-sa ing mode, especially i he engine s a us is O , signalling ha he ehicle is no in ope a ion. This allows he RSU o manage powe consump ion e ec i ely by educing he equency o da a ansmissions du ing low a ic condi ions. Mo eo e , he uel le el can p o ide insigh s in o he ehicle's ope a ional s a us; ehicles wi h low uel le els migh be mo e likely o s op o ha e engine issues, necessi a ing imely communica ion o assis ance. By inco po a ing hese a ibu es, he ensemble lea ning model can be e p edic powe needs and adjus communica ion p o ocols, ul ima ely op imizing ba e y u iliza ion in RSUs. 4.6.1.3 Memo y Space Occupa ion The memo y space occupa ion is signi ican ly a ec ed by ea u es such as mileage, las se ice da e, and ABS s a us. Vehicles wi h highe mileage migh ha e mo e ope a ional da a ha needs o be p ocessed and s o ed, which can s ain he RSU’s memo y esou ces. The las se ice da e helps assess ehicle main enance and ope a ional heal h, allowing RSUs o p io i ize da a caching o ehicles ha a e mo e likely o need eal- ime moni o ing. Addi ionally, he ABS s a us (whe he i 's ac i e o inac i e) can indica e po en ial sa e y issues ha equi e immedia e a en ion and communica ion. By analysing hese ea u es, he ensemble model can lea n op imal caching s a egies, e aining c i ical da a while o loading less ele an in o ma ion, he eby managing memo y mo e e icien ly. The dynamic adjus men o memo y alloca ion based on eal- ime a ic condi ions helps p e en memo y o e load and ensu es ha RSUs can handle he in lux o da a e ec i ely. The expe imen al esul s demons a e how he selec ed ehicle a ibu es can be le e aged o enhance he pe o mance o RSUs in e ms o da a ansmission speed, ba e y powe u iliza ion, and memo y space occupa ion. By employing ensemble lea ning echniques ha inco po a e hese ea u es, RSUs can make in o med decisions o op imize hei ope a ions, he eby imp o ing he o e all e iciency and eliabili y o ehicula communica ion ne wo ks. This in eg a ion o da a- d i en insigh s no only suppo s eal- ime communica ion bu also p omo es sus ainabili y and ope a ional longe i y in RSU deploymen s. 4.6.2 Design Requi emen s To e ec i ely implemen he hyb id machine lea ning model o op imizing RSUs in Vehicula Ad-hoc Ne wo ks (VANETs) using Py hon, se e al design equi emen s mus be es ablished. Fi s , he a chi ec u e should inco po a e obus da a handling capabili ies o manage he di e se ehicle a ibu es, such as speed, loca ion, ba e y ol age, and o he ele an ea u es. This necessi a es he use o lib a ies such as Pandas o da a manipula ion and p e-p ocessing, ensu ing ha da a is cleaned, no malized, and s uc u ed p ope ly o model aining. Addi ionally, amewo ks like NumPy can acili a e e icien nume ical compu a ions, while Ma plo lib o Seabo n can be u ilized o da a isualiza ion, enabling de elope s o analyse da a ends and co ela ions isually. The implemen a ion should also include a modula design ha allows o easy upda es and main enance o indi idual componen s, such as da a collec ion, model aining, and p edic ion. Fu he mo e, he model mus le e age ad anced machine lea ning lib a ies, such as Sciki -lea n and Tenso Flow, o build, ain, and e alua e he ensemble lea ning algo i hms. This in ol es designing a pipeline ha in eg a es a ious models, including Ridge Reg ession, Decision T ees, and Random Fo es s, o o m a cohesi e ensemble. The use o echniques like c oss- alida ion will be c i ical o model e alua ion, ensu ing ha p edic ions a e eliable and gene alizable. The design should also conside scalabili y, allowing he sys em o handle inc easing amoun s o ehicula da a e icien ly. Finally, he implemen a ion mus include pe o mance me ics o assess he model's accu acy, la ency, powe e iciency, and memo y u iliza ion, p o iding a comp ehensi e amewo k o ongoing e alua ion and imp o emen o he RSU esou ce managemen s a egy. 4.7 METRICS 4.7.1 Da a T ansmission Speed Fo mula ion: 𝑇ℎ𝑟𝑜𝑢𝑔ℎ𝑝𝑢𝑡 = 𝐷/𝑇 Eq: (12) Whe e:  D = To al amoun o da a ansmi ed (in bi s)  T = Time aken o he ansmission (in seconds) Th oughpu is a key pe o mance me ic ha indica es how much da a is success ully ansmi ed Jou nal o Theo e ical and Applied In o ma ion Technology 31s May 2025. Vol.103. No.10 © Li le Lion Scien i ic ISSN: 1992-8645 www.ja i .o g E-ISSN: 1817-3195 4180 o e a ne wo k in a speci ied amoun o ime. This o mula ion allows us o quan i y he e iciency o he RSU in managing da a communica ion unde a ying a ic condi ions. A highe h oughpu sugges s ha he RSU can handle mo e ehicles and da a packe s simul aneously, essen ial o main aining pe o mance in high-densi y scena ios. Addi ionally, la ency can be modelled as: 𝐿𝑎𝑡𝑒𝑛𝑐𝑦 =   Whe e:  𝑇= To al ime aken o a esponse (in milliseconds)  N = Numbe o eques s sen This indica es he a e age delay expe ienced pe eques , highligh ing he esponsi eness o he RSU in eal- ime communica ion. 4.7.2 Ba e y Powe U iliza ion Fo mula ion: 𝐸𝑛𝑒𝑟𝑔𝑦 𝐶𝑜𝑛𝑠𝑢𝑚𝑝𝑡𝑖𝑜𝑛 = 𝑃 ⋅𝑡 Eq: (13) Whe e:  P = Powe consump ion (in Wa s)  = Time o ope a ion (in hou s) Explana ion: This o mula ion calcula es he o al ene gy consumed by he RSU du ing i s ope a ion. I p o ides insigh in o how e icien ly he RSU uses powe unde a ious ope a ional condi ions. Lowe ene gy consump ion du ing peak imes indica es e ec i e ene gy managemen s a egies. To exp ess ba e y li e, we can use: 𝐵𝑎𝑡𝑡𝑒𝑟𝑦 𝐿𝑖𝑓𝑒 =   Eq: (14) Whe e:  𝐸= To al ene gy capaci y o he ba e y (in Wa -hou s)  P = A e age powe consump ion (in Wa s) This o mula ion helps es ima e how long he RSU can ope a e on a ull ba e y cha ge, c ucial o ensu ing con inuous se ice. 4.7.3 Memo y Space Occupa ion Fo mula ion: 𝑀𝑒𝑚𝑜𝑟𝑦 𝑈𝑡𝑖𝑙𝑖𝑧𝑎𝑡𝑖𝑜𝑛 𝑅𝑎𝑡𝑒 =   ×100 Eq: (15) Whe e:  𝑀 = Amoun o memo y cu en ly used (in MB)  𝑀 = To al memo y a ailable (in MB) Explana ion: The memo y u iliza ion a e quan i ies how much o he a ailable memo y is being ac i ely used, exp essed as a pe cen age. A highe u iliza ion a e indica es ha he RSU is s o ing a signi ican amoun o da a, which could lead o po en ial memo y o e load i no managed p ope ly. Fo da a e en ion e iciency, we can conside : 𝐷𝑎𝑡𝑎 𝑅𝑒𝑡𝑒𝑛𝑡𝑖𝑜𝑛 𝑅𝑎𝑡𝑒 =   ×100 Eq: (16) Whe e:  𝐷 = Amoun o da a success ully e ie ed (in MB)  𝐷 = To al amoun o da a s o ed (in MB) This o mula ion e alua es how e icien ly he RSU can access s o ed da a, which is c i ical o main aining pe o mance du ing peak a ic pe iods. 5. RESULTS AND DISCUSSION 5.1 Da a Design The Da a Design sec ion de ines he s uc u e and ini ial s a e o ehicle da a h ough he Vehicle class which is u ilized om he link. (h ps://a chi e.ics.uci.edu/da ase /415/ds c+ ehicl e+communica ions ). The ehicle alues and p ame e s a e conside ed om he OBU ea u es and o such each ehicle ins ance is ini ialized wi h a unique iden i ie ( ehicle_id), and i s a ibu es a e andomly gene a ed o simula e eal-wo ld condi ions. This includes aspec s such as speed, loca ion, uel le el, engine s a us, ba e y ol age, and a ious o he me ics ele an o ehicle pe o mance and condi ion. The use o andomness ensu es a di e se da ase , which can be aluable o es ing and analysis. Jou nal o Theo e ical and Applied In o ma ion Technology 31s May 2025. Vol.103. No.10 © Li le Lion Scien i ic ISSN: 1992-8645 www.ja i .o g E-ISSN: 1817-3195 4181 The gene a e_ini ial_da a me hod is esponsible o popula ing he ehicle a ibu es wi h andom alues wi hin speci ied anges, simula ing he a iabili y in ac ual ehicle da a. Fo example, speed is se be ween 0 and 120 km/h, while loca ion is de ined wi h la i ude and longi ude alues ha span he globe. Each ehicle's s a e can change o e ime, ep esen ed h ough he gene a e_da a me hod, which e eshes he a ibu es, allowing o ongoing simula ion o he ehicle's ope a ional cha ac e is ics. 5.2 Da a Communica ion The Da a Communica ion sec ion implemen s a ehicle- o-in as uc u e communica ion model ia he V2XCommunica ion class. I u ilizes socke s o enable eal- ime da a ans e be ween ehicles and a cen al p ocessing uni (in his case, a Road Side Uni , o RSU). The se e lis ens o incoming ehicle da a in a UDP o ma , which is e icien o b oadcas ing messages o mul iple ecipien s wi hou es ablishing a di ec connec ion. Upon ecei ing da a, i decodes he JSON o ma , which is ligh weigh and easy o pa se, allowing o s uc u ed communica ion. Each ehicle's da a is s o ed in a lis o u he p ocessing. The send_da a me hod, hough no ully de ailed in his con ex , would allow he RSU o send da a back o ehicles, enhancing bidi ec ional communica ion capabili ies. 5.3 Da a P ocessing The Da a P ocessing segmen is encapsula ed in he Da aP ocessingUni class. I s p ima y ole is o manage he ehicle da a s o ed in a Da aF ame (a powe ul da a s uc u e om he Pandas lib a y). This sec ion p o ides me hods o eading da a om a CSV ile, which would ypically be gene a ed by he RSU a e collec ing ehicle da a o e ime. The ead_ om_cs me hod includes e o handling o manage scena ios whe e he ile migh no exis o when unexpec ed issues occu du ing eading. This ensu es obus ness in he da a managemen p ocess. Al hough he class ini ially has me hods o appending da a and e u ning i , hey a e commen ed ou , indica ing a po en ial a ea o expansion based on u he equi emen s. The main ocus he e is o acili a e he o ganiza ion and manipula ion o ehicle da a o subsequen analysis. Fig.3: Rep esen ing he o iginal da ase om he UCI websi e Jou nal o Theo e ical and Applied In o ma ion Technology 31s May 2025. Vol.103. No.10 © Li le Lion Scien i ic ISSN: 1992-8645 www.ja i .o g E-ISSN: 1817-3195 4182 Fig.4: Rep esen ing he o iginal da ase om he UCI websi e The da ase in depic ed in Fig.3. and Fig.4. con ains 10,000 ows and 16 columns, whe e each ow ep esen s eleme y and ope a ional da a o an indi idual ehicle. The ehicle_id column uniquely iden i ies each ehicle, while speed cap u es he ehicle's cu en eloci y. The loca ion column p o ides GPS coo dina es in he o m o la i ude and longi ude, indica ing he ehicle's p ecise geog aphic posi ion. Fuel_le el eco ds he emaining uel in he ehicle as a pe cen age, and engine_s a us deno es whe he he engine is cu en ly unning ("On") o no ("O "). Ba e y_ ol age ep esen s he cu en ol age le el o he ehicle's ba e y, and i e_p essu e eco ds he p essu e in he i es, ypically in uni s such as PSI. Oil_ empe a u e gi es he engine oil's empe a u e in deg ees Celsius, which is essen ial o moni o ing engine pe o mance. The mileage column shows he o al dis ance a elled by he ehicle in i s li e ime, while b ake_s a us indica es he condi ion o he b aking sys em, wi h alues such as "Failu e," "No mal," o "Wa ning." The headligh _s a us speci ies whe he he headligh s a e cu en ly "On" o "O ." The empe a u e_inside column e lec s he in e io o cabin empe a u e, and gps_accu acy measu es he p ecision o he loca ion da a p o ided by he ehicle’s GPS sys em. Las _se ice_da e acks he mos ecen main enance o se ice da e o he ehicle, c ucial o ensu ing op imal pe o mance. Wheel_alignmen _s a us eco ds whe he he ehicle's wheel alignmen is in p ope condi ion ("No mal") o equi es adjus men . Finally, abs_s a us e e s o he ope a ional s a us o he ehicle's An i-lock B aking Sys em (ABS), a key sa e y ea u e ha helps main ain con ol du ing b aking. This da ase is aluable o eal- ime ehicle moni o ing, diagnos ics, and p edic i e main enance, making i pa icula ly ele an in con ex s such as lee managemen o ehicula sa e y sys ems. Fig.5: Rep esen ing he con e ed da ase wi h balanced o m Jou nal o Theo e ical and Applied In o ma ion Technology 31s May 2025. Vol.103. No.10 © Li le Lion Scien i ic ISSN: 1992-8645 www.ja i .o g E-ISSN: 1817-3195 4183 The da ase in Fig.5. e ains he same columns as balanced, including ehicle_id, speed, uel_le el, engine_s a us, ba e y_ ol age, i e_p essu e, oil_ empe a u e, mileage, b ake_s a us, headligh _s a us, and empe a u e_inside. Howe e , i he goal is o gene a e a mo e balanced e sion o he da ase , pa icula ly when dealing wi h imbalanced ca ego ical a iables (such as b ake_s a us o engine_s a us), echniques like SMOTE (Syn he ic Mino i y O e -sampling Technique) can be applied. SMOTE is used o add ess class imbalances by syn he ically gene a ing new da a poin s o he mino i y classes wi hou simply duplica ing da a. He e's how SMOTE would be applied o gene a e a balanced da ase in he con ex o ehicle eleme y da a: Iden i ying Imbalanced Fea u es: Fi s , ca ego ical ea u es such as b ake_s a us o engine_s a us, which could ha e imbalances (e.g., a majo i y o ehicles ha ing "No mal" b ake s a us, while ew ha e "Failu e" o "Wa ning"), would be iden i ied. SMOTE is ypically applied o such ca ego ical labels whe e he mino i y classes need o e sampling. Gene a ing Syn he ic Da a: Fo each unde ep esen ed class, SMOTE gene a es syn he ic ins ances by in e pola ing be ween exis ing ins ances o ha class. Fo example, in he case o b ake_s a us, SMOTE would iden i y ehicles wi h a s a us o "Failu e" o "Wa ning" ( he mino i y classes) and gene a e new syn he ic da a poin s by c ea ing combina ions o exis ing ehicles in hose classes. These syn he ic poin s main ain he ela ionships be ween ea u es like speed, ba e y_ ol age, i e_p essu e, and mileage, ensu ing ha he gene a ed da a esembles ealis ic scena ios. Balanced Da ase : Once SMOTE is applied, he da ase will ha e app oxima ely equal ep esen a ion ac oss he classes in ea u es like b ake_s a us. Fo ins ance, he numbe o ehicles wi h a "Failu e" s a us will inc ease h ough syn he ic da a gene a ion un il i ma ches he coun o he mo e common "No mal" class. This esul s in a mo e balanced da ase wi hou losing he ela ionships be ween a iables such as speed, oil empe a u e, and ba e y ol age. Main aining Fea u e Dis ibu ion: SMOTE ensu es ha he syn he ic da a poin s espec he dis ibu ion o con inuous a iables like uel_le el o oil_ empe a u e. The new ins ances will ha e alues ha a e ealis ic and lie wi hin he ea u e space de ined by he o iginal da ase , ensu ing ha he ela ionships be ween hese a iables emain consis en wi h he o iginal da a. In his con ex , applying SMOTE can enhance he da ase 's balance, especially o ehicle s a us ca ego ies ha migh impac he eliabili y o p edic i e models (e.g., p edic ing main enance needs o ailu es). 5.4 Da a Visualiza ion In his sec ion o da a isualisa ion, he p oposed design wi h IMFT model impa s he di e en columns mapping on di e en scale plo s o ensu e he co ec ep esen a ion o bi a ia e analysis is obse ed. This design imp o ises he eal changes on he da a o p edic he uel cases based on he memo y u iliza ions and da a ansmission speed wi h SNS pai plo in Fig.6. Fig.6: Rep esen ing he SNS pai plo o he da ase Jou nal o Theo e ical and Applied In o ma ion Technology 31s May 2025. Vol.103. No.10 © Li le Lion Scien i ic ISSN: 1992-8645 www.ja i .o g E-ISSN: 1817-3195 4184 5.5 Da a Analysis Fig.7: Rep esen ing The S em Plo Fo The P oposed Design Wi h Fuel And Ene gy Consump ion Pa ame e s In he Da a Analysis block, he Ridge Reg ession class is de ined o pe o m p edic i e modelling on he p ocessed ehicle da a. Ridge eg ession, a egula iza ion echnique, is chosen o p e en o e i ing while es ima ing he ela ionship be ween mul iple independen a iables ( ea u es) and a dependen a iable ( uel le el, in his case).The i ing p ocess in ol es ini ializing weigh s and biases, upda ing hem i e a i ely based on he g adien s o he loss unc ion, and applying L2 egula iza ion o con ol complexi y. A e aining he model, i p edic s uel le els based on he es da a. An adjus men phase ollows, whe e p edic ions ha de ia e signi ican ly om ac ual alues a e co ec ed i e a i ely o imp o e accu acy. The mean pe cen age e o is calcula ed o e alua e he model's pe o mance, p o iding insigh s in o i s p edic i e capabili y. The esul s a e isualized using a plo o compa e p edic ed and ac ual uel le els, which aids in in e p e ing he model's e ec i eness and iden i ying any disc epancies in Fig.7. In his app oach, he design alues a e calcula ed based on he cu en L2 egula iza ion o each ype o me ic calcula ion o he p oposed Ridge classi ie indica ing he bes means squa e e o as abula ed in Table-II and Table-III. Simila ly, he s em plo is designed o ep esen he exac and p edic ed alues. As om he Fig.6. he design depic s 0.5% o e o obse ed as indica ed abo e. Fig.8: Rep esen ing Ba Plo Fo Da a T ansmission And La ency Fo RSU Design. Jou nal o Theo e ical and Applied In o ma ion Technology 31s May 2025. Vol.103. No.10 © Li le Lion Scien i ic ISSN: 1992-8645 www.ja i .o g E-ISSN: 1817-3195 4185 Fig.9: Rep esen ing line plo o Da a T ansmission and La ency o RSU design. The Fig.8. and Fig.9. p esen s wo plo s illus a ing he da a ansmission speed and la ency o 100 ehicles in a Vehicula Ad-hoc Ne wo k (VANET) en i onmen , likely in e ac ing wi h a Road Side Uni (RSU). The le plo shows he ansmission speed in megabi s pe second (Mbps) ac oss ehicles, wi h he X-axis ep esen ing he Vehicle ID (0 o 100) and he Y-axis indica ing ansmission speed. The speeds luc ua e be ween 24 Mbps and 28 Mbps, wi h an a e age a ound 26 Mbps, sugges ing ha ac o s such as ehicle dis ance om he RSU, signal in e e ence, o ne wo k load could be a ec ing he ansmission pe o mance. The a iabili y obse ed ac oss ehicles unde sco es he need o obus da a handling mechanisms in he VANET design o ensu e consis en communica ion quali y. The igh plo displays he la ency expe ienced by each ehicle, measu ed in nanoseconds (ns). He e, he X-axis ep esen s he Vehicle ID, and he Y- axis deno es la ency, which anges om 900 ns o 1075 ns. This a iabili y in la ency, wi h mos ehicles expe iencing delays be ween 950 ns and 1025 ns, may be a ibu ed o ac o s such as ne wo k conges ion, ehicle mobili y, and a ia ions in signal s eng h. The pe o mance esul s shown in bo h plo s a e c i ical o VANET design conside a ions. In pa icula , op imizing he placemen and ope a ion o RSUs and adop ing dynamic communica ion p o ocols a e essen ial o mi iga ing ansmission delays and main aining consis en speeds, ensu ing he o e all eliabili y and e iciency o he ne wo k. 5.6 In eg a ion & Execu ion The inal sec ion ocuses on In eg a ion & Execu ion, which b ings oge he he p e iously de ined componen s. The main unc ion o ches a es he en i e low, s a ing wi h he simula ion o ehicles gene a ing da a and sending i o he RSU. Mul iple h eads a e u ilized o simula e he concu en ope a ion o se e al ehicles, enhancing ealism in da a gene a ion. I also includes pe iodic da a sa ing unc ionali y, ensu ing ha he collec ed da a is p ese ed and can be analyzed la e . A e simula ing he ehicles, he Da a P ocessing Uni eads he sa ed CSV ile in o a Da aF ame, p epa ing i o analysis. Finally, he main unc ion calls he analysis module o i he eg ession model on he p ocessed da a, culmina ing in he p esen a ion o esul s. 5.7 Tabula ions The Table-I p esen s pe o mance me ics o a Ridge eg ession model applied o p edic alues in a uel and ene gy sys em, po en ially o ehicle- ela ed da a such as uel consump ion, ene gy usage, o o he ope a ional me ics. Each ow co esponds o a es case, wi h columns showing he p edic ed alue, ac ual alue, and key e o me ics like Absolu e E o , Mean Squa ed E o (MSE), Roo Mean Squa ed E o (RMSE), and Mean Absolu e E o (MAE). Jou nal o Theo e ical and Applied In o ma ion Technology 31s May 2025. Vol.103. No.10 © Li le Lion Scien i ic ISSN: 1992-8645 www.ja i .o g E-ISSN: 1817-3195 4186 Table-I: Rep esen ing he o e all pe o mance me ics o he p oposed design using Ridge Classi ica ion modelling and Reg ession analysis INDEX FINAL PREDICTED VALUE REAL VALUE ABSOLUTE ERROR MSE RMSE MAE R2 SCORE 0 28.8678 29 0.1322 0.0181 0.1346 0.2434 0.99 1 73.5674 74 0.4326 0.1876 0.4337 0.2434 0.99 2 19.8851 20 0.1149 0.0132 0.1149 0.2434 0.99 3 6.5234 7 0.4766 0.2275 0.4765 0.2434 0.99 4 48.6267 49 0.3733 0.1396 0.3733 0.2434 0.99 5 37.7212 38 0.2788 0.0778 0.2788 0.2434 0.99 6 37.7934 38 0.2066 0.0427 0.2066 0.2434 0.99 7 85.6408 86 0.3592 0.128 0.3592 0.2434 0.99 8 19.9482 20 0.0518 0.0027 0.0518 0.2434 0.99 9 70.5659 71 0.4341 0.1885 0.4341 0.2434 0.99 Table-II: Rep esen ing he o e all E o pe o mance o he exis ing and p oposed algo i hms The inal column, R² Sco e, indica es how well he model i s he da a. Wi h a consis en ly high R² sco e o 0.99 ac oss he able, he model explains 99% o he a iance in he eal alues, demons a ing an excellen i . The low alues o Absolu e E o , RMSE, and MAE ac oss all obse a ions sugges he Ridge eg ession model is making highly accu a e p edic ions, wi h minimal de ia ion om ac ual alues. This le el o p ecision is c ucial o eal- ime applica ions like op imizing uel consump ion o managing ene gy e iciency in ehicles, whe e e en small p edic ion e o s can lead o signi ican ope a ional gains. In his case, an in a-in e mod il e design o he Ridge eg esso likely in ol es a dual-laye ed app oach o op imize uel and ene gy solu ions. In a- il e ing e e s o p edic ions made wi hin indi idual ehicle sys ems ( uel le els, ba e y ol age, e c.), op imizing ene gy usage o uel e iciency on a pe - ehicle basis. In e - il e ing, on he o he hand, agg ega es da a om mul iple ehicles o op imize la ge sys ems such as uel dis ibu ion ne wo ks o sha ed cha ging in as uc u e. Ridge eg ession’s abili y o handle mul icollinea i y ensu es ha he model emains obus , gene alizing well e en in he p esence o noisy da a. I s egula iza ion ea u e p e en s o e i ing, esul ing in eliable p edic ions ac oss di e en con ex s. This balance be ween indi idual and sys em-wide p edic ions, combined wi h low e o a es and a high R² sco e, makes he Ridge eg ession model an e ec i e ool o imp o ing uel e iciency and ene gy op imiza ion ac oss bo h ehicle and lee -le el applica ions. MODEL MEAN SQUARED ERROR (MSE) ROOT MEAN SQUARED ERROR (RMSE) MEAN ABSOLUTE ERROR (MAE) R² SCORE RANDOM FOREST [2] (HYPER TUNNED) 870.56 29.51 25.55 -0.0006 RIDGE (HYPER TUNNED) [3] 867.21 29.45 25.47 0.0033 LINEAR REGRESSION (HYPER TUNNED) [6] 867.21 29.45 25.47 0.0033 LASSO (HYPER TUNNED) [7] 867.34 29.45 25.48 0.0031 ELASTIC NET (HYPER TUNNED) [8] 867.34 29.45 25.47 0.0031 PROPOSED (IMFT) RIDGE (ERROR TUNNED) 0.0996 0.2876 0.2434 0.99 Jou nal o Theo e ical and Applied In o ma ion Technology 31s May 2025. Vol.103. No.10 © Li le Lion Scien i ic ISSN: 1992-8645 www.ja i .o g E-ISSN: 1817-3195 4187 Table-III: Rep esen ing he o e all me ic pe o mance o he p oposed algo i hm a e p edic ion VEHICLE ID FUEL LEVEL TRANSMISSION SPEED (MBPS) LATENCY (NS) 237 80 40.11 638.19 238 93 46.3 552.86 239 21 10.43 2453.51 240 4 1.89 13573.56 241 15 7.58 3377.98 242 45 22.44 1140.59 243 92 45.95 557.09 244 22 10.98 2331.87 245 28 14.04 1822.77 246 41 20.46 1251.18 247 19 9.2 2784.01 248 87 43.55 587.77 249 16 8.04 3182.53 250 2 0.98 26174.64 251 97 48.39 529.02 252 5 2.54 10097.91 253 1 0.54 47590.41 254 90 44.95 569.48 255 20 9.98 2565.56 256 84 41.91 610.88 257 18 8.98 2850.1 258 65 32.34 791.47 259 28 14.04 1823.14 260 60 29.91 856.03 The Table-III p esen s da a on ehicles ele an o Vehicle- o-Vehicle (V2V) communica ion, ocusing on a ibu es such as uel le el, ansmission speed (in Mbps), and la ency (in nanoseconds). Each ehicle is iden i ied by a unique ID, wi h a ying uel le els impac ing hei ope a ional ange and communica ion capabili ies. T ansmission speeds di e signi ican ly among ehicles, di ec ly a ec ing he e iciency o da a exchange. High la ency alues indica e po en ial connec i i y issues ha could hinde eal- ime applica ions, making i essen ial o he design o Roadside Uni s (RSUs) o conside hese ac o s o op imal communica ion. Using a p oposed idge eg ession model, insigh s can be de i ed om his da ase o p edic ansmission speeds and la ency based on exis ing a ibu es. This p edic i e analysis aids in unde s anding how a iables in e ac , acili a ing be e decision-making o RSU placemen and esou ce alloca ion. By iden i ying ehicles a isk o high la ency, s a egies can be de eloped o enhance connec i i y and eliabili y in V2V ne wo ks. O e all, le e aging his da a can lead o imp o ed sa e y and e iciency in ehicula communica ions. The pe o mance me ics o a ious models used o p edic ing ehicle communica ion a ibu es highligh signi ican di e ences in hei e ec i eness. In Table-II, he Random Fo es model, e en when hype - uned, exhibi s a mean squa ed e o (MSE) o 870.56 and a nega i e R² sco e o -0.0006, indica ing poo p edic i e pe o mance. In con as , idge eg ession, linea eg ession, lasso, and elas ic ne models all achie e simila MSEs a ound 867.21 o 867.34 wi h R² sco es close o ze o (0.0033 o 0.0031), sugges ing limi ed explana o y powe in he con ex o he da a.In s a k con as , he p oposed idge model (e o uned) demons a es excep ional