[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-83921-en":3,"doc-seo-83921-105":29,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":13,"seo_description":14,"update_tm":27,"read_time":28},83921,1099514068035,"Ezra","https://ap-avatar.wpscdn.com/davatar_276721f389ce27ea32af1340a28f341c",8,"Research & Report","Optimal Base Station Placement for Beyond 5G Networks with Non-Convex Topology","This paper investigates optimal millimeter-wave (mmWave) base station placement in a realistic U-shaped environment with non-convex topology. The placement task is formulated as an NP-hard problem because both the constraints and objectives are non-convex, combining sum-rate maximization with max–min fairness that is additionally nonsmooth. A Markov Decision Process (MDP) is adopted, and two deep reinforcement learning methods are proposed: grid-based DQN and partitioned multi-space DDPG. Results show full coverage and a Jain index of 0.99, with partitioned DDPG achieving better performance than DQN at lower complexity.","Optimal Base Station Placement for Beyond 5G Networks with Non-Convex Topology  \nMohamed Shalma, Amr Mansour, and Ahmed El-Mahdy  \nFaculty of Information Engineering and Technology  \nThe German University in Cairo  \nCairo, Egypt  \n[mohamed.hamed@guc.edu.eg](mohamed.hamed@guc.edu.eg) ; [amr.mansour@student.guc.edu.eg](amr.mansour@student.guc.edu.eg) ; [ahmed.elmahdy@guc.edu.eg](ahmed.elmahdy@guc.edu.eg)  \narXiv :2607 .05210v1 [ ee ss . SP] 6 Jul 2026  \nAbstract—This paper investigates the optimal placement of a millimeter-wave (mmWave) base station (BS) within a realistic U-shaped environment with non-convex topology. The problem is challenging and NP-hard due to the non-convex topology and the non-convex objective functions which are the sum-rate maximization and max–min fairness, the latter being additionally nonsmooth. To address this challenge, the BS placement is formulated as a Markov Decision Process (MDP). Then, we propose two deep reinforcement learning (DRL) techniques: First, the deployment area is discretized into a grid and optimized using a Deep QNetwork (DQN). Second, the U-shaped region is partitioned into continuous subspaces, where a Deep Deterministic Policy Gradient (DDPG) agent is dedicated to each subspace then the best BS placement is selected among partitions. Results demonstrate that optimal placement achieves full coverage and yields a Jain index of 0.99. Furthermore, the proposed partitioned multi-space DDPG achieves better solution than DQN with lower complexity.  \nIndex Terms—base station (BS), placement, optimization, deep reinforcement learning (DRL), DDPG, DQN.  \nI. INTRODUCTION  \nThe transition toward Beyond 5G (B5G) and 6G wireless architectures is driven by the demand for ultra-high data rates, millisecond-level latency, and massive device connectivity. To meet these requirements, network operators are increasingly leveraging several technologies [1], [2], including millimeterwave (mmWave) frequencies, which offer expansive bandwidths but present significant propagation challenges. Unlike sub-6 GHz signals, mmWave transmissions are highly susceptible to severe path loss and physical blockages from urban infrastructure. Consequently, the strategic placement of base stations (BSs) becomes a critical determinant of network performance, as even minor spatial deviations can lead to significant signal degradation.  \nTraditional network planning often relies on static mathematical solvers or exhaustive search algorithms. However, in complex environments with non-convex constraints and high-dimensional search spaces, these methods become computationally prohibitive and often provide worse solutions compared to recent advancements in Artificial Intelligence (AI) [3] . By modeling infrastructure deployment as a sequential decision-making problem, Deep Reinforcement Learning (DRL), in particular, enables an intelligent agent to navigate complex topological constraints and learn optimal spatial  \nmappings through interaction with a simulated environment [4] .  \nThis paper investigates the optimal placement of a mmWave base station within the C-Buildings complex at the German University in Cairo (GUC) . By formulating the placement challenge as a Markov Decision Process (MDP), we utilize a Deep Q-Network (DQN) to optimize for two distinct objectives: Max-Min Fairness for equitable user coverage and Sum-Rate Maximization for aggregate system capacity. Our work contributes to the development of self-organizing B5G networks by demonstrating the efficacy of AI-driven spatial optimization in realistic, campus-scale environments.  \nThe contributions of this paper are summarized as:  \n• We investigate the optimal placement of the BS in practical urban scenario using the mmWaves frequency for B5G networks. The practical permitted vicinity for BS placement is non-convex U-shaped topology leading to an NP hard non-convex optimization problem.  \n• Beside the non-convex topology, We consider two nonconvex objective fu","cbCaittvTfKhEqxY","https://ap.wps.com/l/cbCaittvTfKhEqxY","pdf",2882868,1,6,"English","en",105,"# Introduction\n# Literature Review","[{\"question\":\"Why is the base station placement problem NP-hard in this work?\",\"answer\":\"Because the environment has a non-convex U-shaped topology and the optimization objectives are also non-convex, including a max–min fairness term that is additionally nonsmooth.\"},{\"question\":\"How is the placement problem modeled for the proposed learning methods?\",\"answer\":\"The paper formulates the base station placement as a Markov Decision Process (MDP), enabling sequential decision-making through interaction with a simulated environment.\"},{\"question\":\"What are the two deep reinforcement learning approaches proposed?\",\"answer\":\"One approach discretizes the deployment area into a grid and uses a Deep Q-Network (DQN). The second partitions the U-shaped region into continuous subspaces and assigns a Deep Deterministic Policy Gradient (DDPG) agent per partition, selecting the best placement among them.\"}]",1784191457,15,{"code":4,"msg":30,"data":31},"ok",{"site_id":24,"language":23,"slug":32,"title":13,"keywords":33,"description":14,"schema_data":34,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":27},"optimal-base-station-placement-for-beyond-5g-networks-with-non-convex-topology","",{"@graph":35,"@context":85},[36,53,68],{"@type":37,"itemListElement":38},"BreadcrumbList",[39,43,47,50],{"item":40,"name":41,"@type":42,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":44,"name":45,"@type":42,"position":46},"https://docshare.wps.com/document/","Document",2,{"item":48,"name":12,"@type":42,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":42,"position":52},"https://docshare.wps.com/document/optimal-base-station-placement-for-beyond-5g-networks-with-non-convex-topology/83921/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":23,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":40,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-17","2026-07-16",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"Why is the base station placement problem NP-hard in this work?","Question",{"text":75,"@type":76},"Because the environment has a non-convex U-shaped topology and the optimization objectives are also non-convex, including a max–min fairness term that is additionally nonsmooth.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How is the placement problem modeled for the proposed learning methods?",{"text":80,"@type":76},"The paper formulates the base station placement as a Markov Decision Process (MDP), enabling sequential decision-making through interaction with a simulated environment.",{"name":82,"@type":73,"acceptedAnswer":83},"What are the two deep reinforcement learning approaches proposed?",{"text":84,"@type":76},"One approach discretizes the deployment area into a grid and uses a Deep Q-Network (DQN). The second partitions the U-shaped region into continuous subspaces and assigns a Deep Deterministic Policy Gradient (DDPG) agent per partition, selecting the best placement among them.","https://schema.org",{"og:url":51,"og:type":87,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":89,"canonical":51},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,114,119,122,127,130,134],{"id":20,"doc_module":4,"doc_module_name":45,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":46,"doc_module":4,"doc_module_name":45,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":45,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":45,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":21,"doc_module":4,"doc_module_name":45,"category_name":111,"show_sort_weight":112,"slug":113},"Technology",50,"technology",{"id":115,"doc_module":4,"doc_module_name":45,"category_name":116,"show_sort_weight":117,"slug":118},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":120,"slug":121},30,"research-report",{"id":123,"doc_module":4,"doc_module_name":45,"category_name":124,"show_sort_weight":125,"slug":126},9,"Religion & Spirituality",20,"religion-spirituality",{"id":125,"doc_module":4,"doc_module_name":45,"category_name":128,"show_sort_weight":125,"slug":129},"World Cup","world-cup",{"id":131,"doc_module":4,"doc_module_name":45,"category_name":132,"show_sort_weight":131,"slug":133},10,"Lifestyle","lifestyle",{"id":135,"doc_module":4,"doc_module_name":45,"category_name":136,"show_sort_weight":106,"slug":137},19,"General","general"]