Indonesian J our nal of Electrical Engineering and Computer Science V ol. 43, No. 3, September 2026, pp. 749 ∼ 761 ISSN: 2502-4752, DOI: 10.11591/ijeecs.v43.i3.pp749-761 ❒ 749 Seamless U A V integration: a framew ork f or netw ork slicing as a ser vice in 5G and bey ond Bouzid T ar ek, Nour eddine Chaib, Mohamed Lahcen Bensaad Laboratoire d’Informatique et de Math ´ ematiques, Uni v ersit ´ e Amar T elidji de Laghouat, Laghouat, Algeria Article Inf o Article history: Recei v ed Mar 11, 2026 Re vised Jul 15, 2026 Accepted Jul 23, 2026 K eyw ords: 5G B5G Netw ork slicing Service frame w ork U A V -assisted netw ork slicing Unmanned aerial v ehicles ABSTRA CT Unmanned aerial v ehicles (U A Vs) ha v e demonstrated remarkable v ersatility across monitoring, deli v ery , and data collection applications. Ho we v er , the gro wing comple xity of netw ork architectures with 5G and Be yond 5G (B5G) necessitates inno v ati v e solutions lik e netw ork slicing, which f aces challenges in radio access netw ork (RAN) ef cienc y and optimization under higher frequen- cies. This paper introduces a no v el frame w ork for seamless U A V -assisted net- w ork slicing as a service. By inte grating U A Vs directly into the netw ork infras- tructure, the frame w ork enables on-demand deli v ery of tailored netw ork slices to end users, addressing the limitations of traditional terrestrial approaches while enhancing ef cienc y and scalability . The proposed frame w ork allo ws service pro viders and netw ork administrators to deli v er slices more ef fecti v ely , while specically tackling high-frequenc y RAN challenges by deplo ying U A Vs as close-proximity relays to users. A k e y contrib ution is complete compatibility and interoperability with e xisting cellular infrastructure, ensuring seamless inte- gration. The paper presents critical foundational concepts and a literature re vie w , follo wed by t he proposed frame w ork’ s architecture, layers, and methodology . Comprehensi v e tests and comparisons e v aluate the frame w ork’ s performance, concluding with a discussion of applications and potential enhancements. This is an open access article under the CC BY -SA license . Corresponding A uthor: Bouzid T arek Laboratoire d’Informatique et de Math ´ ematiques, Uni v ersit ´ e Amar T elidji de Laghouat Laghouat, Algeria Email: t.bouzid@lagh-uni v .dz 1. INTR ODUCTION Future wireless netw orks must support di v ers e services under highly dynami c operating condi tions, moti v ating e xible resource management mechanisms such as netw ork s licing. In this conte xt, unmanned aerial v ehicles (U A Vs) of fer on-demand aerial support that enhances netw ork adaptability and impro v es communica- tion performance, particularly in challenging deplo yment scenarios [1]. Their rapid mobility also mak es them well suited to mitig ating the propag ation limitations associated with high frequenc y communications in 6G systems, thereby helping sustain reliable quality of service (QoS) for demanding applications [2]. Building upon the demonstrated capabilities of U A V -assisted netw orking across multiple c o m muni- cation scenarios [3]-[5], we incorporate a dedicated aerial service layer that complements terrestrial cellular infrastructures. This cooperati v e architecture impro v es resource utilization while maintaining seamless inter - operability between aerial and ground netw ork components. This frame w ork becomes a compelling proposition for netw ork pro viders in the b usiness conte xt. U A Vs’ superior throughputs and dynamic deplo yment capabilities position them as catalysts for meeting di- J ournal homepage: http://ijeecs.iaescor e .com Evaluation Warning : The document was created with Spire.PDF for Python.
750 ❒ ISSN: 2502-4752 v erse use cases. This is particularly appealing for service pro viders aiming to dif ferentiate their of ferings, capitalize on emer ging applications, and tailor solutions to v arying user requirements. Indeed, the frame w ork’ s compatibility with e xisting infrastructures increases its feasibility as a practi cal solution for enhancing service portfolios and maintaining a competit i v e edge. Combining the agility and e xibility of U A Vs with the features of netw ork slicing allo ws users to get the best service possible. As a result, this paper will present a 5G/B5G compatible service frame w ork for U A V -assisted netw ork slicing. Due to the increased demands and e xpecta tions of netw ork de vices, 5G m ade radical changes to the pre vious generations. These upgrades include netw ork slicing features, softw are dened netw orking (SDN), functions virtualization (NFV), and more. The additions are necessary to address lar ge netw ork requirement demands. Fundamentally , 5G comprises tw o primary portions: the Core (which is similar to e v olv ed pack et core (EPC) of 4G), the radio access netw ork (RAN), another component connecting these portions i s the trans- port netw ork. Figure 1 depicts the 5G system architecture, sho wcas ing the infrastructure and logical layers [6]. A detailed description of these portions and ho w the y constitute 5G cellular netw ork is gi v en belo w [6]. Figure 1. 5G architecture and deplo yment representation As pre viously stated, this component mirrors the foundational purpose of 4G’ s EPC, yet it di v er ges signicantly by shifti ng to w ard an NFV -dri v en virtualized deplo yment. Because the core architecture is con- structed out of virtualized netw ork functions (VNFs), it g ains the e xibility required to seamlessly manage netw ork slicing and resource distrib ution. This core domain orchestrates se v eral primary services called man- agement functions, such as session (SMF), user plane (UPF), and access and mobility (AMF) [7]. By operating through a collaborati v e information-sharing ecosystem, these functions allo w e xternal architectures (including the frame w ork de v eloped in this w ork) to easily interf ace with them and e xtract rele v ant data. Interconnecting the Core with the RAN represents the primary objecti v e of the transport stratum. W ithin this domain, SDN is instrumental in e x ecuting traf c prioritization and o v erall data o w management [8]. While it is rarely highlighted as a headline component of the netw ork, the transport’ s ability to control o ws and lter pack ets renders it an enabler for netw ork slicing. The e v olution of fth-generation mobile systems has shifted radio access design to w ard e xible and scalable architectures capable of supporting services with widely dif ferent communication requirements. This e v olution is reected in the generation partnership project (3GPP) ne w radio (NR) standard, which operates o v er tw o spectrum domains [9]: frequenc y range 1 (FR1), spanning 0.41 GHz to 7.1 GHz, and frequenc y range 2 (FR2), co v ering between 24.25 GHz and 71 GHz. The lar ge bandwidth a v ailable in FR2 enables high-capacity wireless services b ut also introduces signicant propag ation challenges, as millimeter -w a v e t ransmissions e xperience se v ere attenuation, are easily obstructed, and pro vide limited co v erage distances. These characteristics moti v ate the use of complementary communication platforms that can be deplo yed where v er x ed infrastructure alone cannot pro vide adequate radio performance. Indonesian J Elec Eng & Comp Sci, V ol. 43, No. 3, September 2026: 749–761 Evaluation Warning : The document was created with Spire.PDF for Python.
Indonesian J Elec Eng & Comp Sci ISSN: 2502-4752 ❒ 751 Indeed, U A V mobility allo ws the netw ork to respond rapidly to changing communication condit ions without requiring permanent infrastructure e xpansion. Such inte gration is supported by the modular or g ani- zation of the 5G RAN, where radio functionality is distrib uted among Units, namely the cent ralized (CUs), distrib uted (DUs), and radio units (R Us). Because these functions are separated, additional radio endpoints can be incorporated with minimal disruption to the e xisting deplo yment. User equipment (UE) are main user de vices of the netw ork and should support the higher frequenci es. Recent de vices are already com p a tible with the 5G standard, massi v e multiple input multiple output (MIMO) and higher frequencites. 5G encompasses a di v erse set of end users (outside typical phones or tablets), which include v ehicles, drones, smart equipment, and internet of things (IoT) de vices. A fundamental capability of 5G/6G sys tems is the ability to partition a common ph ysical netw ork into se v eral isolated logical en vironments, each optimized for a particular cate gory of applications or performance objecti v es. This concept, kno wn as netw ork slicing, enables heterogeneous services to share the same netw ork without interfering with eachother . Its implementation is lar gely enabled by NFV and SDN, which pro vide the e xibility required to dynamically instantiate and manage virtual netw ork components to create slices. 3GPP dened three primary slice cate gories to support the major service scenarios [10]: − Enhanced mobile BroadBand (eMBB): tar gets intensi v e bandwidth demands by deli v ering high-capacity connecti vity , enabling applications that consume lar ge v olumes of data, including immersi v e multimedia and cloud-centric services. U A Vs can supplement terrestrial infrast ructure by pro viding additional co v erage and capacity in traf c hotspots, thereby maintaining high throughput and impro ving user e xperience. − Ultra reliable lo w latenc y communica tion (URLLC): supports scenarios where communication delays and transmission f ailures must be m inimized, making it suitable for time-sensiti v e operations such as intelligent transportation, industrial control, and critical healthcare systems. By dynamically positioning themselv es near users, U A Vs can reduce communication delays and impro v e link reliabili ty , helping satisfy stringent QoS requirements. − Massi v e machine type communication (mMTC): f acilitates the ef cient connection of enormous popula- tions of lo w-po wer de vices that e xchange relati v ely small amounts of data, pro viding the communication foundation for lar ge-scale IoT ecosystems and perv asi v e sensing applications. U A Vs can e xtend connecti v- ity to dense or remote IoT deplo yments, ef ciently aggre g ating traf c from numerous de vices. Gro wing research interest has demonstrated that U A V communication platforms can signicantly im - pro v e netw ork performance across a broad range of deplo yment scenarios [3]-[5]. Moreo v er , U A V -assisted netw ork slicing g ained much popularity in recent years. Indeed, U A Vs as a service ha v e been emplo yed in man y netw ork assisting scenarios [11]-[13]. U A Vs can pro vide better cell-edge connecti vity for netw ork users [14], pro vide edge capabilities [15], and data collection services [16], [17]. U A Vs can also assist in deli v ering netw ork slices. F or instance, Y ang et al. [18] use a user location prediction model to impro v e URLLC slice deli v ery by U A Vs. Whereas in [19], authors impro v e mMTC deli v ery using deep reinforcement learning to maximizing resource and ener gy ef cienc y in a U A V -assisted netw ork. Furthermore, fe w frame w orks ha v e been proposed to e xploit U A Vs in netw ork slicing. Y uan and Muntean [20] test the viability of U A V -enhanced QoS by proposing a proof of concept frame w ork. Sk ondras et al. [21], the authors propose an SDN-enabled slicing frame w ork for v ehicular communication which le v er - ages U A V relays to dynamically allocate communication resources and enhance service performance in 5G. The frame w ork proposed in this paper e xtends our pre vious publications [22], addresses all major netw ork slices (eMBB, mMTC, and URLLC), and presents U A Vs as a service t o enhance netw ork slice deli v ery while being highly compatible and compliant with the 5G and B5G architecture. Moreo v er , upgrades are also proposed to enhance the frame w ork, including char ging stations which can be deplo yed on a per -cell basis (the base station (BS)) and using custom drone types for each slice type [23]. The remainder of this article is structured as follo ws. Section 2 present s our proposed frame w ork and its components before detailing the frame w ork methodology and architecture. Section 3 pro vides tests and e v aluations of the proposed service. Finally , potential optimizations and applications are discussed before concluding. 2. METHOD The frame w ork introduced in this w ork can be fully compatible with the current and ne xt generation of cellular architecture with minima l to no changes to the netw ork foundation. the proposed frame w ork consists of Seamless U A V inte gr ation: a fr ame work for network slicing as a service in 5G and be yond (Bouzid T ar ek) Evaluation Warning : The document was created with Spire.PDF for Python.
752 ❒ ISSN: 2502-4752 multiple interconnected components operating within the netw ork, as depicted in Figure 2. Its implementation relies on virtualized core functionalities, softw are-based e x ecution o v er standard-purpose serv ers rather than proprietary netw orking de vices, and the dynamic congurability of fered by se gmented netw ork domains i,e ,. Core, RAN, and transport. The proposed frame w ork components and properties are cited belo w: Globa l   ar chit ectur e The  R AN R u s ( R ad io   Un its ) Dis trib u t ed   Un its C en tr aliz ed   Un its Th e C or e UR L L C  U A V m M T C  U A V e M BB  U A V A l l   p u r p os e   d r on e s Da t a   f l ow Da t a   (pa y l oad)   f l ow U A V s  /  b a t t e r y   s t a tu s Figure 2. Ov ervie w of the frame w ork and its components embedded into the end-to-end netw ork architecture 2.1. Framew ork components 2.1.1. Or chestrator (Hierar chical coordination and ser vice intelligence lay er) The proposed coordination frame w ork acts as an intelligent bridge between high-le v el b usiness poli- cies and real-time netw ork operations by continuously analyzing netw ork conditions and managing both aerial and terrestrial resources. Hierarchical control architecture. Aligning with the decentralized nature of 5G radio access net w orks, orchestration is distrib uted across edge locations rather than a remote cloud. Local controllers (distrib uted ph ys- ically within the CUs) manage their respecti v e geographic domains with reduced latenc y while coordinating through a logically unied management layer to preserv e system-wide consistenc y . Operational w orko w and functionality . The orchestor continuously e x ecutes a closed control loop composed of v e major tasks: − Netw ork observ ation: the distrib uted instances continuously g athers real-time telemetry from U A Vs, radio access nodes, char ging f acilities, and user sessions to establish an accurate, system-wide netw ork state. − Conte xt analysis: proacti v ely e v aluates resource a v ailability , traf c demand, and U A V hardw are conditions to anticipate netw ork changes before service de gradation occurs. This a llo ws the system to either command the U A V to rechar ge or e v en preempti v ely deplo y U A Vs where necessary . − Decision generation: selects optimal management actions—such as deplo ying or relocating U A Vs, initiating char ging c ycles, reallocating virtual resources, or e x ecuting hando v ers—based on operator objecti v es. − Command dissemination: translates selected actions into control-plane signaling instructions distrib uted to infrastructure entities, utilizing continuous feedback loops with netw ork observ ation and analysis to v erify task e x ecution. − Polic y compliance and b usiness optimization: enforces centrally dened strate gic policies (e.g., QoS tar - gets, comm ercial rules) at the edge while reporting operational statistics back to the global high layer for long-term polic y renement. By separat ing strate gic high global s up e rvision from localized edge e x ecut ion, t his hierarchical model signicantly enhances netw ork scalability and o v erall operational rob ustness while maintaining b usiness polic y control for administrators. Indonesian J Elec Eng & Comp Sci, V ol. 43, No. 3, September 2026: 749–761 Evaluation Warning : The document was created with Spire.PDF for Python.
Indonesian J Elec Eng & Comp Sci ISSN: 2502-4752 ❒ 753 2.1.2. Char ging infrastructur e and ener gy support f or U A Vs Reliable operation of the proposed service depe n ds on a distrib uted char ging infrastructure. Each char ging f acility serv es as an ener gy replenishment point, enabling drones to restore their batteries for future missions. Besides supplying po wer , these f acilit ies function as communication g ate w ays between grounded drones w aiting for assignment and the orchestration layer which can send acti v ation instructions. The aim of pro viding seamless po wer transfer without human interactions is a major eld, often rely- ing on rechar ging frame w orks or battery stati on s [24]. Additionally , automatic battery replacement in v entions are proposed, some of which can replace batteries on the y . This ef cienc y and autonomy can impro v e service response times and eliminate do wntime. 2.1.3. U A Vs W ithin the proposed frame w ork, U A Vs operate as adapti v e aerial netw ork nodes pro viding on-demand communication services. The drone eet is geographically distrib uted across the co v erage re gions of multiple BSs, thus, U A Vs remain close to areas where additional netw ork capacity or co v erage may be required. Drone communication. A U A V remains in standby while, or after char ging until it rece i v es an act i- v ation request from the orchestrator . F ollo wing deplo yment, the drone establishes a wireless connection with its cell station, becoming part of the radio access layer . During operation, the U A V forw ards user -plane traf c and e xchanges control-plane information for mobility management and coordination. By positioning the drone near end users, the system enhances link quality and impro v es performance [1]. Specic drone types. The proposed frame w ork does not assume identical drone types. Instead, the y may be equipped with dif ferent communication modules and onboard resources according to their role or tar get slice. Some platforms can incorporate edge-processing and storage capabilities to e x ecute latenc y-sensiti v e services locally , whereas others may focus primarily on radio co v erage enhancement and support netw ork slicing requirements. 2.2. Framew ork model Figure 3 presents the interactions among main parts of the frame w ork architecture within the netw ork. The distrib uted orchestration i nstances coordinate re gional resources (including drones and stations) under the supervision of a logically centralized management application. Moreo v er , Char ging platforms maintain U A V readiness and support the deplo yment of inacti v e drones when required. Si mplifi ed  basic ar chit ectur e  / descrip tion  la y er s  deplo ymen t ce n tr al   c ontr ol  and   admi n is tr a tio n Mai n  Or ch es tr a t or Or ch es tr a t or  ins t ances Or ch e s tr a t o r  in s t anc e s  d i vide d   and   loca t e d   in   diff er ent  ce ll  ar eas E ach  ins t ance  ov er sees   its  r espectiv e BS  user s a nd  r eso ur ces   inclu din g  the  ch ar ging  pla tf orms a nd   dr ones The   C or e U A V s   in fr as tructu r e  la y er f u n ct i on s Use  the  c or e and  its  vir tual   ne tw ork  fu nc tions   (VNF s)  t o  oper a t e, g a ther s ess ion,  slice  and   u s er  in f orm a tio n R ech arg e  pl a t f or ms mMT C   use r s De vices  c an  c onn ect  dir ectly  t o  their  BSs  or  t o  a dr one if  the ser vice is   enabled a nd   r equ es t ed Di f f er en t  sli c e  us er s Figure 3. Outlining the multilayer architecture of the frame w ork inte gration with the netw ork Figure 4 depicts a typical cell distrib ution with the BSs, U A Vs, users, and char ging stations. The CUs’ and DUs’ are not depicted as their infrastructure plac ement v aries depending on the service pro vider and tar get area. Seamless U A V inte gr ation: a fr ame work for network slicing as a service in 5G and be yond (Bouzid T ar ek) Evaluation Warning : The document was created with Spire.PDF for Python.
754 ❒ ISSN: 2502-4752 Figure 4. System model illustration with tw o adjacent cells Modern 5G infrastructures already pro vide the architectural foundations required by the proposed frame w ork including virtualized netw ork functions, softw are-dened control, alongside RAN components lik e CUs. As a result, drone platforms can be i nte grated as dynamically managed resources to meet the requirements of dif ferent netw ork users. 2.3. Demanding the slicing ser vice Figure 5 presents a sequence diagram sho wcasing interactions between the frame w ork and netw ork components. The sequence diagram summarizes the system beha vior from the user connection and service request to the user e xploitation of the service. Figure 5. User connection and service request sequence diagram 2.4. Decision framew ork and strategy Control actions are deri v ed from real-time netw ork information e xchanged with the 5G Core. Indeed, the orchestrator continuously e v al uates the system states before determining the appropriate management ac- tions based on the main decision criteria gi v en belo w: − Slice priority algorithm: indeed, some slices, such as URLLC, are critical and delay-sensiti v e. As a result, the planner considers these requirements. F or instance, the planner can prioritize URLLC traf c o v er eMBB slices, lea ving non-delay sensiti v e mMTC traf c for last. − U A V states: the planner should al w ays consider U A V states and battery le v els. This is critical to maintain service quality and reduce do wntime. Rather than maintaining e xisting associations, the planner dynami- cally redistrib utes user sessions among eligible drones to pre v ent resource congestion based on their states and localization. − Potential congestion and scalability: the planner al w ays ensures that users are distrib uted equally across the a v ailable U A Vs. This will reduce U A V o v erload, increasing scalability . As for the orchestrator , the instances are already distrib uted, which ensures scalability and rob ustness Indonesian J Elec Eng & Comp Sci, V ol. 43, No. 3, September 2026: 749–761 Evaluation Warning : The document was created with Spire.PDF for Python.
Indonesian J Elec Eng & Comp Sci ISSN: 2502-4752 ❒ 755 − Resource management: the planner should al w ays consider the a v ailable resources. Thes e include stations, U A Vs, and edge capabilities if a v ailable. More specically , if the service of fers edge resources on U A Vs, the planner should consider their utilization and distrib ution. 2.5. Ener gy-awar e U A V r eplacement strategy T o maintain uninterrupted service, the frame w ork monitors the ener gy le v el replaces U A Vs before battery depletion af fects netw ork performance. Once a drone reaches the predened ener gy threshold, it is recalled to the nearest char ging station. The o v erall replacement procedure is illustrated in Figure 6. Figure 6. U A V replacement o wchart In essence, replacement candidate U A Vs must support the required slice and possess suf cient c apac- ity to accommodate the transferred users. If no suitable single drone is a v ailable, user sessions are redistrib uted between multiple operational drones to preserv e connection continuity . 2.6. Adapti v e session quality management The local orchestrator continuously analyzes acti v e connections to detect an y performance de gra da- tion. When a user’ s communication no longer meets the requirements of its assigned slice (such as latenc y or bandwidth), the session is migrated to a better -suited drone to maintain the user’ s requested le v el of QoS. 3. RESUL TS AND DISCUSSION The e v aluations are conducted using a simulation en vironment that reproduces a dense urban 5G deplo yment and compares the proposed frame w ork with a con v entional BS-only architecture. W ireless prop- ag ation is modeled according to the 3GPP Urban Macro (UMa) specication [25], incorporating practical chan- nel characteristics including antenna heights, transmission po wer , carrier frequenc y , and penetration ef fects. The terrestrial BS operates at a height of 25 m with a transmit po wer of 40 dBm, whereas user equip- ment is positioned at 1.5 meters. Drones are deplo yed between 50 and 150 m abo v e ground le v el to pro vide aerial netw ork support. T raf c is generated by up to 50,000 users dis trib uted o v er a 1 × 1 km 2 service area. User requests are dynamically assigned to one of the three supported netw ork slices, with peak bandwidth demands reaching 500 Mbps. The drone eet size v aries according to the e v aluated scenario, ranging from 3 in the baseline conguration to 51 in high-capacity e xperiments. The simulation (drone and BS) settings and slice-specic ph ysical-layer parameters are presented in T able 1 and T able 2, respecti v ely . T able 1. Adopted drone and BS settings Properties BS eMBB drone mMTC drone URLLC drone Drone cate gory N/A He xacopter Fix ed-W ing Hybrid Drone v elocity N/A 6 m/s 4 m/s 5 m/s T ransmit po wer 40 dBm 30 dBm 22 dBm 26 dBm Carrier frequenc y range 2 to 5 GHz 5 GHz 0.5 to 0.8 GHz 1.8 to 2.5 GHz Height range 25 met ers 80 to 120 meters 30 to 60 meters 80 meters Battery capacity N/A 16,000 mAh 10,000 mAh 11,000 mAh T able 2. Assumed settings per netw ork slice Adopted parameter eMBB slice mMTC slice URLLC slice Numerology µ 1 0 2 Modulation scheme [26] QAM256 QAM64 (Lo wSE) QAM64 (Lo wSE) Subcarrier spacing 2 µ · 15 k H z 30 kHz 15 kHz 60 kHz P ack et size 4 Kbits 1 Kbi ts 3 Kbits Channel width 100 MHz 5 MHz 100 MHz Maximum throughput per user 500 M bps 25 Mbps 200 Mbps Seamless U A V inte gr ation: a fr ame work for network slicing as a service in 5G and be yond (Bouzid T ar ek) Evaluation Warning : The document was created with Spire.PDF for Python.
756 ❒ ISSN: 2502-4752 3.1. P erf ormance e v aluation against a BS F or the frame w ork to be a viable, it must pro vide similar or better performance compared to the traditional netw orks, i.e ., using a BS. T o this end, we rst e v aluate the proposed U A V system ag ainst a BS system before continuing to other frame w ork service performance metrics. 3.1.1. Number of connections Figure 7 demonstrates that the number of supported user connections depends on user load. While both systems deli v er similar performance at lo w traf c le v els, increasing user density e xposes the capacity lim- itations of the BS deplo ym ent, particularly for high requirement slices. Consequently , eMBB and URLLC e x- perience a progressi v e reduction in satised users, whereas mMTC remains comparati v ely unaf fected because of its users’ lo wer resource demands. The proposed U A V -assisted frame w ork e xhibits a dif ferent scalability trend seen in Figure 7. By distrib uti ng drone access points throughout the co v erage area, communication links remain short er with higher line of sight (LoS) chance, making radio conditions impro v e and enabling a greater proportion of users to remain connected e v en under hea vy netw ork load. Figure 7. Comparing a v erage user connections between the proposed frame w ork and a BS-only system 3.1.2. Sent trafc Indeed, high data rates are a main pri ority for current netw ork users, especially for the eMBB slice. Therefore, the e v aluation depicted in Figure 8 compares the total traf c sent under the proposed system vs the BS system within the same time-frame. As observ ed in Figure 8, the BS has a lo wer traf c when increasing user numbers since higher data rates require better links. As U A Vs can pro vide better links to the users, we see that the performance is held e v en with increased user numbers and a total of 10 U A Vs per slice. As a result, the U A V service can guarantee higher throughputs. 0 2 0 0 4 0 0 6 0 0 8 0 0 1 00 0 1 2 0 0 20 50 1 0 0 5 0 0 1 0 0 0 2 0 0 0 5 0 0 0 1 0 0 0 0 2 0 0 0 0 5 0 0 0 0 S en t m es s ag es  ( mill io ns) Num b e r  of   u ser s T o tal   sen t  tr affi c  fo r   u ser s  co n n ected   to   th e  B S  v s  30  U A Vs Sen t  m e ssag es  ( BS  c o n n ec tio n ) Sen t  m e ssag es  ( UA V  co n n ec tio n ) Figure 8. Comparing between a BS system and the proposed U A V service in the amount of sent traf c 3.2. Ev aluating the ser vice perf ormance 3.2.1. System r esponse time Figure 9 presents the user w aiting beha vior for U A V service connections in terms of the a v erage w aiting time and the percentage of users w aiting for service. Indeed, as the service relies on U A Vs to reach the users, some delay concerns may arise. As a result, this aspect is thoroughly e xamined. The e v aluations depicted in Figure 9(a) sho wcase the a v eraged response times in each slice for users that w ait for a connection. Indonesian J Elec Eng & Comp Sci, V ol. 43, No. 3, September 2026: 749–761 Evaluation Warning : The document was created with Spire.PDF for Python.
Indonesian J Elec Eng & Comp Sci ISSN: 2502-4752 ❒ 757 W e observ e that e v en with one U A V per slice, the a v erage w aiting time is relati v ely lo w at 11 seconds and decreases as more U A Vs can be used in the cell. Ho we v er , most users connect instantly after requesting the service, and only a small percentage of users e xperience the w aiting times sho wn in Figure 9(a). The percentage of users that might w ait for the service are less than 6% as depicted in Figure 9(b). Most of these users are the rst that r equire the service as no U A Vs will be deplo yed. This percentage quickly diminishes as more U A Vs are allocated in the cell to serv e users. W e also observ e that mMTC users ha v e the lo wers w aiting percentage, approaching 0% w aiting users when deplo ying more than 8 mMTC U A Vs. This lo w percentage is due to their long range communications, as the rst mMTC users can connect as soon as the U A V launches. 0 2 4 6 8 10 12 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51 A v er ag e  w ait  tim e  ( s) Num b e r  of   alloc a t e d   U A V s U s e r   a v e r a g e   w a i t i n g   t i m e   f o r   U A V   c o n n e c t i o n   v s   n u m b e r   o f   U A V s eM BB  wai t  t im e U RL LC w ai t   t im e m MT C w ai t   t im e (a) 0 1 2 3 4 5 6 7 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51 P e r c e n t ag e   o f Use r s  ( % ) Num b e r  of   alloc a t e d   U A V s U s e r s   w a i t i n g   f o r   s e r v i c e   c o n n e c t i o n   v s   n u m b e r   o f   U A V s   p e r   c e l l eM BB  us er s U RL LC u s e rs m MT C us er s (b) Figure 9. User w aiting beha vior for U A V service connection (a) Comparing a v erage user w aiting times for U A V service connection and (b) Percentage of users that w ait for the U A V service connection vs number of a v ailable U A Vs Furthermore, the e v aluation depict ed in Figure 10 highlights the login time of the rst and last U A V for each slice. W e observ e that across all slices, the U A Vs start launching in less than 5 seconds from the simulation start. Ho we v er , the last U A Vs login later on, to handle increasing user requests. 0 5 10 15 20 25 30 20 50 1 0 0 5 0 0 1 0 0 0 2 0 0 0 5 0 0 0 1 0 0 0 0 2 0 0 0 0 5 0 0 0 0 U A V  lo gin   tim e  ( s) Num b e r  of   c e ll  u ser s F i r s t   a n d   l a s t   s l i c e   U A V   l o g i n   t i m e   v s   n u m b e r   o f   u s e r s   Fis t  e MB B UA V lo gin La s t  e MBB  UAV log in Fis t  U RL LC UAV  lo gin La s t  U RLLC   U AV lo gi n Fis t  m M T C   U AV lo g in La s t  m MT C U A V lo gin Figure 10. Comparing the rst and last U A V login time for each slice vs number of users Seamless U A V inte gr ation: a fr ame work for network slicing as a service in 5G and be yond (Bouzid T ar ek) Evaluation Warning : The document was created with Spire.PDF for Python.
758 ❒ ISSN: 2502-4752 3.2.2. Number of ser vice r equests T o see the ef fect of increasing service requests, v arious request probabilities are simulated. Since the proposed service is optional, the users may opt to use the U A V service or to just connect to the nearest BS. In Figure 11, the total users in each cell ha v e the probability of 20%, 50%, and 100% to request the service after logging in. F or instance, in the case of 50%, half the connected users will request the service. Whereas, the other half of users will connect to the BS normally . W e notice in Figure 11 that increasing the number of user requests does not ne g ati v ely ef fect perfor - mance, ho we v er , it may require more U A Vs, especially at 100%. Indeed, the performance of the U A V system does k eep up with the increasing requests, and it can reach full capacity if U A Vs are suf cient. 0 1 0 0 0 2 0 0 0 3 0 0 0 4 0 0 0 5 00 0 6 0 0 0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51 Ser v ed   u ser s Num b e r  of   U A V s  p e r  c e ll E f f e c t   o f   s e r v i c e   r e q u e s t s   o n   U A V   s e r v i c e   p e r f o r m a n c e 20 %  s e rv ice  r e q u e s t 50 %  s e rv ice  r e q u e s t 10 0%  s e rv ice  r e q u e s t 0 10 20 30 40 50 60 70 80 90 1 0 0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51 Ser v ed   user s  pe r c en t ag e % Num b e r  of   U A V s  p e r  c e ll E f f e c t   o f   s e r v i c e   r e q u e s t s   o n   U A V   s e r v i c e   p e r f o r m a n c e 20 %  s e rv ice  r e q u e s t 50 %  s e rv ice  r e q u e s t 10 0%  s e rv ice  r e q u e s t Figure 11. Comparing user requests for the U A V service and its ef fect on performance 3.2.3. Number of U A Vs Indeed, the proposed frame w ork will be af fected by the number of a v ailable U A Vs in each cell. The follo wing e v aluation studies this af fect by v arying U A V numbers and observing user connections for each netw ork slice. The e v aluation presented in Figure 12 depicts a total of 50000 users requesting the service. W e clearly observ e that mMTC users reach the highest connection rates with fe wer U A Vs compared to other slices. This pattern is due to the lo w requirements of the mMTC slice. On the contrary , eMBB users are the most demanding, and require more U A Vs compared to other slices. 0 5 0 0 1 0 0 0 1 50 0 2 0 0 0 2 5 0 0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51 N u m b er  of   c onn ec t ed   u ser s Num b e r  of   alloc a t e d   U A V s A v e r a g e   c o n n e c t i o n   r a t e s   v s   n u m b e r   o f   U A V s   p e r   c e l l C on n ect e d   e M B B   u sers C on n ect e d   UR LLC   u ser s C on n ecte d   mM TC u sers Figure 12. Comparing the ef fect of increasing service U A Vs on the number of connections for slice users 3.3. Discussion and potential applications 3.3.1. Additions and optimizations Using cro wd-sourced U A Vs with an incenti v e approach. As U A Vs become more widely a v ailable, the number of U A V o wners is increasing. Thi s presents a potential for the netw ork to e xploit cro wd-sourced U A Vs. Owners may be willing to of fer their drones in e xchange for nancial g ains. This also benets the service pro viders by remo ving the cost of U A V purchase and maintenance, pro viding a benet for all parties. Indonesian J Elec Eng & Comp Sci, V ol. 43, No. 3, September 2026: 749–761 Evaluation Warning : The document was created with Spire.PDF for Python.