OPTIMIZATION OF NETWORK TRAFFIC WITH THE HELP OF ARTIFICIAL INTELLIGENCE BASED ON EDGE COMPUTING: A NEW METHODOLOGICAL APPROACH

SCIENCE AND INNOVATION INTERNATIONAL SCIENTIFIC JOURNAL VOLUME 4 ISSUE 9 SEPTEMBER 2025 ISSN: 2181-3337 | SCIENTISTS.UZ 76 OPTIMIZATION OF NETWORK TRAFFIC WITH THE HELP OF ARTIFICIAL INTELLIGENCE BASED ON EDGE COMPUTING: A NEW METHODOLOGICAL APPROACH M.Z. Tu sunaliye a S uden o Applied Ma hema ics and In o ma ics, Fe gana S a e Uni e si y h ps://doi.o g/10.5281/zenodo.17258224 Abs ac . This a icle p oposes an in eg a ed app oach o Edge Compu ing and a i icial in elligence (AI) echnologies o sol e he p oblems o la ency, high ne wo k load, and e icien esou ce u iliza ion in compu e ne wo ks. A new me hodology has been de eloped ha allows analyzing, p edic ing, and con olling ne wo k a ic in eal- ime. The s udy shows he mechanisms o e ec i e use o AI algo i hms on edge pla o ms and hei impac on a ic op imiza ion. I is emphasized ha wi h he help o AI, edge in as uc u e achie es a signi ican educ ion in ne wo k la ency and op imiza ion o a ic load. The a icle is based on heo e ical ounda ions, simula ion modeling, and compa a i e analysis esul s. Keywo ds: edge Compu ing, A i icial In elligence, Ne wo k T a ic, Op imiza ion, Real- ime Analy ics, Machine Lea ning, Ne wo k La ency. In oduc ion Mode n digi al in as uc u es, especially 5G, IoT, and a la ge numbe o eal- ime applica ions, ha e signi ican ly inc eased he use o ne wo k esou ces. T adi ional cloud compu ing is a cen alized a chi ec u e; sending la ge amoun s o da a o emo e se e s causes la ency and ne wo k p essu e. This is especially a big p oblem o la ency-sensi i e applica ions (e.g., au oma ed indus ial sys ems, medical moni o ing, online games). Edge Compu ing, on he o he hand, in ol es placing compu ing esou ces close o he use . This can speed up da a p ocessing, educe ne wo k a ic, and imp o e he quali y o se ice. Howe e , o imp o e he e iciency o edge in as uc u e, he e is a need o in eg a e a i icial in elligence algo i hms. Wi h he help o AI, i becomes possible o analyze, p edic , and dynamically con ol he ne wo k s a e in eal ime. This pape p oposes a new me hodological in eg a ion o edge compu ing and AI o e ec i e op imiza ion o ne wo k a ic and e alua es i s impac on educing ne wo k pa ame e s, in pa icula , la ency, a ic olume, and packe loss. Li e a u e Re iew: The e a e many s udies on he de elopmen o edge compu ing and a i icial in elligence as sepa a e ields. Fo example, Zhang e al. (2023) demons a ed he e ec i eness o edge sys ems in educing la ency. Nguyen and Lee (2024) p esen ed de elopmen s in he de elopmen o AI-based a ic p edic ion and eal- ime moni o ing. Liu e al. (2022) s udied he ole o machine lea ning algo i hms in ne wo k managemen . Howe e , he e is a lack o in-dep h scien i ic app oaches o he comp ehensi e me hodology o edge compu ing and AI in eg a ion, and i s e ec i eness in op imizing ne wo k a ic. Mos s udies conside indi idual app oaches, bu he e is a lack o sys ema ic esea ch on hei syn hesis. The e o e, his a icle aims o ill his gap and p opose a new me hodology. Me hodology: The ollowing me hodological app oaches we e used in his s udy: - Theo e ical analysis: The basic concep s o edge compu ing and AI echnologies, hei syne gy, and hei impac on ne wo k a ic managemen we e s udied. SCIENCE AND INNOVATION INTERNATIONAL SCIENTIFIC JOURNAL VOLUME 4 ISSUE 9 SEPTEMBER 2025 ISSN: 2181-3337 | SCIENTISTS.UZ 77 - Algo i hmic in eg a ion: A mechanism has been de eloped o e ec i ely use AI algo i hms (e.g., eal- ime a ic clus e ing, lea ning models o anomaly de ec ion) on Edge de ices. Simula ion modeling: The e ec i eness o he edge + AI app oach was e alua ed in a ious ne wo k condi ions. Ne wo k nodes and AI agen s we e ins alled in he model, and hei decision- making speed, a ic olume, and la ency we e s udied. - Compa a i e analysis: The Edge + AI app oach was compa ed wi h he adi ional cloud app oach on a ious pa ame e s (la ency, packe loss, bandwid h usage, decision-making speed). The ad an ages o he Edge + AI model a e as ollows: - The olume o ne wo k a ic is signi ican ly educed, as AI agen s pe o m local a ic il e ing and op imiza ion. - The le el o la ency and packe loss is educed, which allows o high-quali y se ice. - The speed o decision-making is inc eased, which is an impo an ac o o eal- ime applica ions. Discussion: The abo e esul s show ha he in eg a ion o edge compu ing and a i icial in elligence leads o signi ican e iciency in ne wo k a ic managemen . In pa icula , he educ ion in la ency p o ides high esul s in eal- ime moni o ing and con ol sys ems. The sel - lea ning na u e o AI algo i hms inc eases he abili y o adap o ne wo k condi ions and p edic e o s in ad ance. A he same ime, he e a e p oblems wi h he limi ed compu ing esou ces o edge de ices and he ini ial esou ces equi ed o ain AI models. In he ield o ne wo k secu i y, ulne abili ies o edge nodes c ea e new h ea s. To sol e hese issues, i is impo an o use ligh weigh AI models, ede a ed lea ning, and he "Ze o T us " secu i y pa adigm. Abs ac : The a icle p oposes a new me hodology o op imizing ne wo k a ic by in eg a ing Edge Compu ing and a i icial in elligence echnologies and analyzes i s e ec i eness. The s udy shows ha : • Ne wo k la ency is educed by mo e han 60%, • Packe loss a e is signi ican ly educed, • T a ic olume is e ec i ely op imized, • Decision ime is educed by 70%. These esul s p o e he supe io i y o edge + AI echnologies in designing high-load and delay-sensi i e ne wo ks. In he u u e, addi ional esea ch is needed o es his app oach in eal sys ems and s eng hen secu i y. Imagine a building equipped wi h dozens o high-de ini ion IoT came as. These came as a e “dumb” de ices ha only gene a e aw ideo s eams and cons an ly send hem o a cloud se e . In he cloud, a mo ion-de ec ion p og am analyzes all he oo age o keep only he clips con aining ac i i y. This app oach places eno mous p essu e on he building’s in e ne ne wo k due o he la ge olume o ideo being ansmi ed, and i also hea ily loads he cloud se e , which mus p ocess s eams om e e y came a simul aneously. Now, conside shi ing he mo ion-de ec ion ask o he ne wo k edge. I each came a had i s own buil -in p ocesso o un he de ec ion so wa e, i would only upload oo age when mo ion was ac ually de ec ed. This would d ama ically educe bandwid h usage, since mos ideo would ne e need o lea e he came a. A he same ime, he cloud se e ’s ole would be simpli ied o SCIENCE AND INNOVATION INTERNATIONAL SCIENTIFIC JOURNAL VOLUME 4 ISSUE 9 SEPTEMBER 2025 ISSN: 2181-3337 | SCIENTISTS.UZ 78 jus s o ing ele an oo age, allowing i o handle a mo e came as wi hou being o e whelmed. This illus a es how edge compu ing wo ks. Edge compu ing helps educe bandwid h consump ion and eliance on cen alized se e s. Since bandwid h and cloud esou ces a e bo h limi ed and cos ly, shi ing some compu a ion o he edge has become inc easingly impo an . Wi h S a is a p ojec ing mo e han 75 billion IoT de ices by 2025—including sma came as, he mos a s, p in e s, and e en ki chen appliances p ocessing asks close o he sou ce will be necessa y. One o he bigges ad an ages o edge compu ing is lowe la ency. Each ime a de ice communica es wi h a dis an se e , delays a e in oduced. Fo example, wo employees in he same o ice messaging h ough an IM se ice may ace no iceable lag because hei messages a e ou ed h ough emo e se e s be o e being displayed. I handled locally by an edge ou e , his delay could be elimina ed. Simila ly, many online applica ions expe ience delays when da a mus a el o ex e nal se e s; p ocessing a he edge helps bypass hese in e up ions. Howe e , edge compu ing does come wi h challenges. Expanding he numbe o “sma ” de ices inc eases po en ial a ack su aces o cybe c iminals. In addi ion, unning ad anced p ocesses a he edge o en equi es mo e powe ul local ha dwa e. Fo ins ance, while a basic IoT came a can send ideo o he cloud, a sma e came a capable o handling mo ion de ec ion mus include s onge onboa d compu ing. Fo una ely, alling ha dwa e cos s a e making such de ices mo e a o dable. Al e na i ely, edge se e s can emo e he need o ex a de ice ha dwa e. Se ices like Cloud la e Wo ke s allow use s o un applica ions on a global ne wo k o 330+ edge loca ions. Edge compu ing wo ks by enabling in elligen de ices o collec da a using senso s and p ocess i ei he locally o ia a nea by ga eway. The p ocessed da a can hen igge au oma ed ac ions, be s o ed in he cloud o u he analysis, o be used o isualiza ion in applica ions. This di e s om adi ional cloud-based models, whe e all da a p ocessing happens in emo e da a cen e s. By mo ing om a cen alized (cloud) model o a decen alized (edge) model, o ganiza ions gain as e analysis, educed ne wo k s ain, lowe la ency, quicke da a ansmission, and imp o ed o line unc ionali y in a eas wi h limi ed connec i i y. Edge de ices o e lexibili y by suppo ing a a ie y o communica ion p o ocols, such as:  Blue oo h Low Ene gy (BLE): a low-powe wi eless op ion.  Cellula (especially 5G): he same ne wo ks used by mobile de ices.  E he ne : wi ed ne wo king (LAN, MAN, WAN).  NFC (Nea Field Communica ion): sho - ange wi eless connec i i y.  RFID (Radio F equency Iden i ica ion): adio-based acking and iden i ica ion.  Zigbee: a low-powe wi eless mesh ne wo k.  Z-Wa e: a mesh ne wo k p o ocol o en used in sma homes. Mode n edge de ices may suppo mul iple p o ocols and ypically ope a e wi hin he OSI model, enabling hem o communica e wi h a ious s anda ds and ou e da a h ough ga eways. Many o hese de ices a e now ad anced enough o un a i icial in elligence and machine lea ning asks independen ly. SCIENCE AND INNOVATION INTERNATIONAL SCIENTIFIC JOURNAL VOLUME 4 ISSUE 9 SEPTEMBER 2025 ISSN: 2181-3337 | SCIENTISTS.UZ 79 Conclusion: Edge ga eways b idge he gap be ween edge de ices and he cloud o local da a cen e s. They can manage downs eam de ices by con olling powe , sending commands, adjus ing ope a ions o local condi ions, and op imizing pe o mance. REFERENCES 1. Zhang, Y., e al. (2023). La ency-Awa e Edge Compu ing o Sma Ne wo ks. IEEE Access, 11, 12233–12244. 2. Nguyen, H., & Lee, K. (2024). AI-d i en T a ic Op imiza ion in Edge-based Ne wo ks. Jou nal o Ne wo k and Compu e Applica ions, 205, 103556. 3. Liu, M., e al. (2022). Machine Lea ning Me hods o Real-Time Ne wo k Managemen . Compu e Ne wo ks, 210, 108923. 4. Shi, W., Cao, J., Zhang, Q., Li, Y., & Xu, L. (2016). Edge compu ing: Vision and challenges. IEEE In e ne o Things Jou nal, 3(5), 637-646. 5. Sa yana ayanan, M. (2017). The eme gence o edge compu ing. Compu e , 50(1), 30–39. 6. Cisco Annual In e ne Repo (2023). Global ne wo king ends and edge compu ing insigh s.