A Novel Airborne Self-Organising Architecture for 5G+ Networks

UWS Academic Portal A novel airborne self organising architecture for 5G+ Networks Shakir, Muhammad Zeeshan; Ahmadi, Hamed; Katzis, Konstantinos Published in: 2017 IEEE 86th Vehicular Technology Conference (VTC Fall) DOI: 10.1109/VTCFall.2017.8288095 Published: 12/02/2018 Document Version Peer reviewed version Link to publication on the UWS Academic Portal Citation for published version (APA): Shakir, M. Z., Ahmadi, H., & Katzis, K. (2018). A novel airborne self organising architecture for 5G+ Networks. In 2017 IEEE 86th Vehicular Technology Conference (VTC Fall) IEEE. https://doi.org/10.1109/VTCFall.2017.8288095 General rights Copyright and moral rights for the publications made accessible in the UWS Academic Portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. Take down policy If you believe that this document breaches copyright please contact [email protected] providing details, and we will remove access to the work immediately and investigate your claim. Download date: 16 Jul 2020 1 A Novel Airborne Self-organising Architecture for 5G+ Networks Hamed Ahmadi, Senior Member, IEEE, Konstantinos Katzis, Senior Member, IEEE, and Muhammad Zeeshan Shakir, Senior Member, IEEE Abstract—Network Flying Platforms (NFPs) such as unmanned aerial vehicles, unmanned balloons or drones flying at low/medium/high altitude can be employed to enhance network coverage and capacity by deploying a swarm of flying platforms that implement novel radio resource management techniques. In this paper, we propose a novel layered architecture where NFPs, of various types and flying at low/medium/high layers in a swarm of flying platforms, are considered as an integrated part of the future cellular networks to inject additional capacity and expand the coverage for exceptional scenarios (sports events, concerts, etc.) and hard-to-reach areas (rural or sparsely populated areas). Successful roll-out of the proposed architecture depends on several factors including, but are not limited to: network optimisation for NFP placement and association, safety operations of NFP for network/equipment security, and reliability for NFP transport and control/signaling mechanisms. In this work, we formulate the optimum placement of NFP at a Lower Layer (LL) by exploiting the airborne Self-organising Network (SON) features. Our initial simulations show the NFP-LL can serve more User Equipment (UE)s using this placement technique. Index Terms—Airborne SON; 5G+ wireless networks; radio access network (RAN); networked flying platforms (NFPs); unmanned aerial vehicle (UAV); drones; low altitude platform (LAP); medium altitude platform (MAP); high altitude platform (HAP); I. I NTRODUCTION The Fifth Generation (5G) networks, which are expected to be rolled out soon after 2020, will support a 1000 times higher average data traffic, 10-100 fold increase in data rate and connected devices [1]. One of the enabling solutions to meet the demand for high data rate is the densification in cellular network by complementing the ultra-dense deployment of low power small base stations (SBSs) with the airborne cellular network and forming a multi-tier heterogeneous network (HetNet) for 5G+ systems. One of the biggest challenges in designing such an airborne cellular network communication system, is to optimally position the flying Base Station (BS) or drone-cell and maintain that position so that the network can benefit the most. Extensive work by numerous research groups investigated how the position of the flying BS or drone-cell can affect the performance of the aerial and terrestrial communication systems as well how its position can play a vital role in the operation of the network. For instance, in [2] authors looked at the effect of different generic mobility models that can be used to characterize the pertinent movements of HAPs H. Ahmadi is with the School of Electrical and Electronic Engineering, University College Dublin, Ireland, Email: [email protected]. K. Katzis is with the Department of Computer Science and Engineering, European University Cyprus, Cyprus, Email: [email protected]. M. Z. Shakir is with the School of Engineering and Computing, University of the West of Scotland, Paisley, Scotland, UK, Email: [email protected]. and their effect on the communications. Furthermore, system operating parameters such as the flight path, operation location and service duration over a particular service area can be of high significance in terms of optimized communications for the users on the ground if the aim is to provide seamless connectivity during handoff between platforms [3]. [4] introduces a genetic Interference-aware Positioning of Aerial Relays (IPAR) algorithm which is able to find suitable positions for the UAVs that maximize the downlink throughput of the cellular network. Similarly, [5] discusses efficient algorithms developed to compute the optimal position of the drone for maximizing the data rate, which are shown to be highly effective via simulations. In [6], authors presents that the optimal altitude is a function of the maximum allowed pathloss and of the statistical parameters of the urban environment, as defined by the International Telecommunication Union. They also present a closed-form formula for predicting the probability of the geometrical line of sight between a Low Altitude Platform (LAP) and a ground receiver. In [7] authors present a 3-D placement problem with the objective of maximizing the revenue, which is measured by the maximum number of users covered by the drone-cell. To solve this problem, they have formulated an equivalent problem which was solved efficiently to find the location and size of the coverage region, and the altitude of the drone-cell. An airbone cellular network comprises of Network Flying Platform (NFP)s of various types including Unmanned Aerial Vehicle (UAV)s, drones, balloons, and high-altitude/mediumaltitude/low-altitude platforms (HAPs/MAPs/LAPs) to expand the cellular coverage and deliver Internet services to remote and dedicated regions where infrastructure is not available and expensive to deploy. However, a major challenge in such networks is how to design and manage Self-organisation in cellular networks which is mostly seen as a result of distributed decision making [8]. In conventional SON, self-configuration, self-optimization and self-healing are considered as main SON functions. An airborne cellular network has a more dynamic nature compared to a fixed cellular network because the position of its elements may change over the time. This may be due to change in demand, weather conditions, coverage requirements, battery limitations and even due to some realtime traffic changes/abnormalities in the network. In such an airborne cellular network each BS i.e., NFP-LL, must be capable of performing a combination of conventional and emerging SON functionalities for airborne cellular network depending on the role of NFPs in the network. The rest of the paper is organised as follows: A novel airborne cellular architecture is introduced in Section II. Section III presents optimisation problem of the proposed airborne SON and its implementation. Section IV presents simulation 2 Fig. 1. Graphical illustration of the hierarchical airborne self organizing architecture where variety of NFPs are flying at different altitudes and are offering a complementary wireless network. results and some discussions on the results. Conclusions are drawn under section V and finally some challenges are summarised in Section VI. II. A IRBORNE S ELF O RGANISING N ETWORKS In an emerging airborne cellular network, the SON functionalities can be realized differently. For example, when a cell outage is detected, neighboring BSs in conveantional SON compensate this outage by increasing their transmission power or changing their antenna tilt (self-healing) while in an airborne SON, NFPs can also effectively reorganise their positions to compensate the outage with lower energy consumption. The new placements can be decided centrally or locally at each NFP element – distributed decision making. deployment are more prone to the associated network security risks including their safe operation and security of information. In our architecture, NFPs in the medium layer are dual role playing i.e., in addition to relaying, medium altitude platforms (MAPs) are performing surveillance to ensure safe and secure operation of the architectures. The surveillance operation includes network monitoring, surveillance scheduling, and decisions to further optimise the network and ensure reliable and secure operation by exploiting the collaboration between the participating NFPs in the medium layer. NFPs in the lower layer are optimally distributed to offer capacity and expand coverage via resource and interference management. It is to note that in the proposed architecture HAPs and MAPs are flying with fixed/known locations, however, LAPs are flying with relatively random or optimally distributed locations and offering coverage or capacity in an optimal way. A. Proposed Systems Architecture: Fig. 1 shows our proposed novel layered architecture where NFPs are distributed in a hierarchal manner such that the LL is responsible for access or fronthaul for small cells, Higher Layer (HL) is responsible for transport network and Medium Layer (ML) is serving as a relay between the two layers. NFPs in the LL are typical low altitude platforms (LAPs) flying at relatively lower altitudes and responsible for network optimisation including NPF placement and association based on resource allocation, interference management, etc. On the other side, NFPs such as UAV, unmanned balloons or drones flying at low/high/medium altitude can be used to enhance network coverage and capacity by deploying a swarm of flying platforms that implement novel radio resource management techniques [9]. In this paper, operating in the HL belong to High Altitude Platform (HAP) category and are responsible for optimising the resources in transport networks for lower layers. NFPs in the ML belong to the Medium Altitude Platform (MAP) category and are responsible for relaying the network between the lower and higher layers in our proposed architecture. Safety and security of the proposed NFPs driven cellular network is of significant importance as such network B. Functionalities of Layers in Proposed Architecture: The general system architecture of our proposed NFP has been divided into three layers of operation. The HL ranging between 15km-25km, the ML ranging between 5km-15km and the LL which is up to 5km height. For each layer, the operating conditions can vary in terms of wind speed, humidity and temperature. This means that certain type of platforms can be operational at a particular layer, that can cope with the flying conditions as well as provide their services which can range from surveillance, broadband connectivity to backhaul communications as described below. 1) HAP: is a stratospheric platform capable of delivering a wide range of services such as mobile cellular communications, broadband wireless access as well as search and rescue services. A HAP is expected to operate at the HL providing Line of Sight (LOS) connectivity over a wide geographical area (30km radius). Such platforms can either be planes or airships, manned or unmanned that can carry a payload of a few kilograms to a few tones and can stay airborne from a few hours to a few years depending on their type, size, power constraints and fuel capacity. A HAP can provide ubiquitous 3 TABLE I S UMMARY OF AERIAL PLATFORMS PARTICIPAITING IN A IRBONE CELLULAR NETWORK . Platform Name AirShip AirPlane AirBaloon AirCopter (S/P/B/C) Manned Unmanned Max Altitude (approximately) Platform length Platform width (Wing Span) Platform weight Range Max Payload Endurance (M/U) (m) (m) (m) (kg) (km) (kg) (hrs) 25 2.7 16 2.26 0.5 40 1900 1.2 2343 100 380 25 0.5 0.37 Tethered 24 0.8 52 Tethered 6 9 34 7.1 10000 270 250 408 487 200 400 6 15 504 48 30 24 117 24 24 26 250 1700 1400 1360 2.27 99.79 250 7000 2700 52 14 8 32 30 24 5 years 6.5 10 years Lower Layer - LAP Amazon Drone Aerovironment Dragon Eye SkyHook (Helikites) Zepellin-NT MD4-1000 (DHL) Skyship 600 (Charly) Desert Star (Helikites) MRI P2006T Protonex C P B S C S B P P U U U M U M U M U 122 150 2286 2600 3000 3050 3352 4200 4250 0.9 7.31 75 1.03 59 10.05 8.7 1.1 5.48 19.5 1.03 15.2 6.7 11.4 8.2 Medium Layer - MAP Schiebel Camcopter S-100 ScanEagle Airlander 10 General Atomics Prowler II FOTROS EADS SDE Eagle 1 Solar Impulse 2 MQ-1 Predator Anka - A Silver Arrow Sniper P P S P P P P P P P U U M U U U M U U U 5486 5944 6100 7600 7600 7620 8534 8839 9144 9145 3.11 1.6 92 5 6.2 9.3 22.4 8.53 8 9.4 1.24 3.1 43 10.75 17 16.6 72 17 17.3 18 110 16 20000 250 Higher Layer - HAP IAI Heron Predator B (MQ-9B) G520 Strto 1 Northrop Grumman Global Hawk Zephyr 6 Aurora Flight Sciences Perseus Stratobus M-55 Geophysica ISIS (Integrated Sensor is Structure) P P P P P P S P S U U M U U U U M U 10000 15000 16000 18000 18288 19812 20000 21000 21500 8.5 11 13.82 14.5 16.6 20 33 39.9 18 21.79 30 37.46 900 2223 3300 6781 30 1936 350 1852 3670 22779 13995 4965 coverage providing backhaul and control/fleet coordination services for other aerial platforms at lower layers. A fleet of HAPs can be used to provide extended communication coverage and redundancy if necessary, while inter-platform communications can be established employing Free Space Optics (FSO). 2) MAP: are aerial platforms operating at the ML that can be used as a relay between a HAP and a LAP. Depending on the operation scenario, current available MAP in the market are mostly UAV with long endurance capabilities as well as manned aerial vehicles. UAV platforms can stay airborne for several hours and are usually destined for military missions. MAP coverage area is expected to be of up to 5km radius. 3) LAP: such as tethered balloons, drones, operate at the LL. Like HAP and MAP, LAP exhibit common features such as LOS communications with favorable radio conditions, while they have the ability to rapidly deploy a fleet of LAP with modular communication payload capabilities. LAP are currently in high demand for public usage and the current market is driving the drone industry in delivering newer and smarter platforms, with higher endurance and greater range, more agile and efficient aerial vehicles. LAP are optimally distributed to offer capacity and expand coverage via resource and interference management. Table 1 lists a number of aerial platforms that can be employed to carry out a mission on a particular layer. In LL, LAP are expected to feature a relatively small size, limited payload and endurance capabilities in minutes or hours while offering rapid deployment operating below 5km height. LAP are expected to operate in organised / scheduled shifts in order to ensure continuity of service. MAP are expected to be able 7.62 100 22.86 137.16 8790 2.9 3757 850 50 1000 2300 1233 1400 1250 Fixed 900 20 1019 Fixed 926 600 180 2000 2000 1000 400 4896 200 to cope with a heavier payload in order to maintain their role as network relays between LL and ML featuring more functionalities than LAP. They are capable of staying aloft for a greater duration of time (hours / days) in order to support the operation of LAP. MAP are expect to be operational between 5-15km height subject to the local aviation regulations of operation. Finally, HAP operating on HL will be of much bigger size operating at a height around 17-20km height. Such platforms will be carrying a much heavier payload that would support Radio Frequency (RF) and FSO communications. It should be able to stay operational for days/months without require refueling. A possible scenario is to use Protonex as LAP that can stay aloft up to 9 hours while the Airlander10 can operate as MAP providing services at about 10km height. Finally, a HAP such as ISIS can stay aloft for a great duration of time providing connectivity to the MAP. III. O PTIMISATION OF A IRBONE SON In our proposed architecture the position of LAPs in the LL is defined centrally and the NFP has the ability to re-organise the LL to achieve its target, which can be capturing as many UEs, maximizing the achievable rate, and/or fairness among UEs. However, optimum placement is an NP-hard problem. NFP-LL placement to capture maximum UEs can be for- 4 of the global solution. Compared to GA and ACO modeling the mixed integer problems is more straightforward in PSO. mulated as: max U N X X n=1 u=1 U X dn,u B. Decentralized decision making: In the proposed architecture we use centralized decision making to define the position of LAP in LL. However, the u=1  architecture has the potential to integrate a distributed decision 1 RSSn,u > RSS−n,u dn,u = ∀u ∈ {1, . . . , U }, making mechanism for positioning the LAP. Game theory is 0 otherwise the most popular techniques used for designing and analyzing where N is the number of LAPs in NFP-LL, U is the number distributed decision making approaches [15]. Different classes of UEs, and dn,u is 1 if the uth UE is served by LAP of game theoretic approaches can be used to model compen, otherwise it will be zero. RSSn,u denotes the received tition, cooperation, and coalition between different players, signal strength of UE u from LAP n. RSS−n,u denotes the LAPs. Game theoretic analysis of the system tells that if strengths of received signal from other nodes of NFP and the there exists a path that leads the competition/cooperation of independent decision makers to an equilibrium. In other words, Macrocells, if exists. The above mixed integer nonlinear problem can be lin- it enables the system designers to avoid objective functions earized using the technique presented in [10] and commercial that lead the system to instability. In an airborne network settings, parameters of the network packages like CPLEX can solve the mixed integer linear problem. Problems with limited number of UEs, and LAPs will change much faster than a terrestrial network. Therefore, serving as hotspots can be solved with exhaustive search. In learning from past experiences is extremely important in both this work we use this technique. However, in more complex centralized and decentralized decision making mechanisms scenarios even the mixed integer linear problem will be too for airborne SON [16]. In decentralized systems, learning complex. Therefore, metaheuristic algorithms can be used to techniques help reaching an equilibrium in fewer iterations. find an efficient sub-optimum solution [11]. We introduce some of these metaheuristic methods. subject to dn,u ≤ 1, ∀n ∈ {1, . . . , N } 70 Airborne SON Fixed placement 60 Number of captured UEs A. Resource allocation using metaheuristics: Popular metaheuristics in radio resource allocation are Genetic Algorithms (GA), Ant Colony Optimization (ACO), and Particle Swarm Optimization (PSO). 1) GA: code the solution into a chromosome (also known as individual) and evaluates the optimality of the solution with a function called fitness function [12]. In this work the fitness function is the sum captured UEs, and the chromosome consists of the position of UAVs. Crossover and mutation are tools of GA for improving the solutions. Crossover combines fragments of two chromosomes and creates a new chromosome which is another solution in the feasible area. Mutation is the tool which is hired for overcoming the local optimums; mutation changes a gene randomly with the hope of reaching better solutions. The algorithm starts with a randomly generated population of chromosomes and at each iteration it creates a generation of offspring using crossover and mutation. The fittest of parent and offspring generations are kept for the next iteration and the rest are discarded. This way GA keep the population size constant. The algorithm stops after the maximum number of iterations is reached. 2) ACO: has been used in radio resource management for network planning and spectrum allocation [13]. The main idea is taken from the way ants find the shortest path to the food using pheromone. The shorter and more popular paths have higher density of pheromone while longer paths will lose pheromone due to the evaporation. In this model each solution will be coded as a path, and at each iteration each ant chooses a path with a probability that is proportional to the amount of pheromone. After a certain number of iterations the most popular will be selected, i.e. the algorithm converges. 3) PSO: models the social behavior of a group of birds [14]. It consists of a swarm of candidate solutions called particles, which explore an n-dimensional hyperspace in search 50 40 30 20 10 0 1 2 3 4 number of LAPs serving the hotspot Fig. 2. Number of UEs captured by airborne SON compared to the fixed placement of LAPs in NFP-LL. The network has 150 UEs where a third of them are in the demand hotspot. Demand hotspot has a radius of 250 m (e.g. outdoor concert area). IV. S IMULATION RESULTS AND DISCUSSIONS We modeled an NFP where the LL provides service to a demand hotspot. The UEs in the system are served by the macrocell and the NFP-LL assists the macrocell by capturing UEs. The NFP optimizes its LL for a given number of LAPs. Due to weather conditions, battery failure or surveillance duty the system may lose an LAP. In this work, we compared an airborne SON system that reorganizes itself with an NFP with fixed LL placement. Our results in Fig. 2 shows that the airborne SON outperforms the fixed placement. Crosslayer optimisation approaches are required to optimise the placement or positioning of NFPs across the layers in proposed airborne SON. 5 V. C HALLENGES AND FUTURE DIRECTIONS Although airborne systems have attracted industry and academia’s attention in the last couple of years, there still exists several challenges and open research directions. Following are some of the important challenges for future airborne SON: A. Standardization Airborne cellular networks are yet to be standardized. The existing networking standards cannot fully address the challenges of airborne networks and proper standards for airborne communication and networking is required. B. Surveillance Airborne cellular network would offer complementary connectivity services to expand the coverage or inject the capacity under some unknown situations, therefore their successful operation would depend on advanced surveillance mechanisms to detect amateur flying platforms and combat to avoid any further disruption in cellular services. C. Information security Secure transmission of information over wireless links is still a challenge for future Internet architecture. In our proposed system, NFPs in all layers have the ability to move and complement the existing network, therefore, the dynamic nature of the airborne cellular network could be an additional challenge to ensure secure coverage expansion or capacity enhancement. D. Ethics and privacy NFPs and swarm of NFPs may face two-fold challenges in order to comply with regulatory issues related to privacy and ethics. NFPs should be able to protect the privacy of the connected users while following the flying ethics as per regulations and avoiding no-flying zone. E. Testbed and verification Various projects in Europe and United States study and test the performance of future Internet and connectivity architecture, resource allocation techniques, waveforms, and integration of future technologies using advanced testbeds [17]. To the best of our knowledge, none of the existing testbed validation and experimentation provide an environment for testing the proposed airborne SON. VI. C ONCLUSIONS In this paper we have presented a novel layered architecture where NFPs, of various types and flying in low/medium/high layers in a swarm of flying platforms, are considered as an integrated part of the future cellular networks to inject additional capacity and expand the coverage for exceptional scenarios and hard-to-reach areas. In our proposed architecture the position of LAPs in the LL is defined centrally and the NFP has the ability to re-organise the LL to achieve its target, which can be capturing as many UEs, maximizing the achievable rate, and/or fairness among UEs. To evaluate the proposed architecture, we compared an airborne SON system that reorganizes itself with an NFP with fixed LL placement. Our results show that the airborne SON outperforms the fixed placement. R EFERENCES [1] A. Osseiran, F. Boccardi, V. Braun, K. Kusume, P. Marsch, M. Maternia, O. Queseth, M. Schellmann, H. Schotten, H. Taoka, H. Tullberg, M. A. Uusitalo, B. Timus, and M. Fallgren, “Scenarios for 5g mobile and wireless communications: the vision of the metis project,” IEEE Communications Magazine, vol. 52, no. 5, pp. 26–35, May 2014. [2] D. Grace, K. Katzis, D. Pearce, and P. Mitchell, “Low-latency mac-layer handoff for a high altitude platform delivering broadband communications,” URSI Radio Science Bulletin, no. 332, pp. 39–49, 3 2010. [3] K. Katzis and D. Grace, “Inter-high altitude platform handoff for communications systems with directional antennas,” URSI Radio Science Bulletin Special Issue on HAPs, 6 2010. [4] S. Rohde, M. Putzke, and C. Wietfeld, “Ad hoc self-healing of ofdma networks using uav-based relays,” Ad Hoc Networks, vol. 11, no. 7, pp. 1893 – 1906, 2013, theory, Algorithms and Applications of Wireless Networked Robotics Recent Advances in Vehicular Communications and Networking. [Online]. Available: http://www.sciencedirect.com/science/article/pii/S157087051200131X [5] X. Li, D. Guo, H. Yin, and G. Wei, “Drone-assisted public safety wireless broadband network,” in 2015 IEEE Wireless Communications and Networking Conference Workshops (WCNCW), March 2015, pp. 323–328. [6] A. Al-Hourani, S. Kandeepan, and S. Lardner, “Optimal lap altitude for maximum coverage,” IEEE Wireless Communications Letters, vol. 3, no. 6, pp. 569–572, Dec 2014. [7] R. I. Bor-Yaliniz, A. El-Keyi, and H. Yanikomeroglu, “Efficient 3-d placement of an aerial base station in next generation cellular networks,” in 2016 IEEE International Conference on Communications (ICC), May 2016, pp. 1–5. [8] O. G. Aliu, A. Imran, M. A. Imran, and B. Evans, “A survey of self organisation in future cellular networks,” IEEE Communications Surveys & Tutorials, vol. 15, no. 1, pp. 336–361, 2013. [9] E. Kalantari, M. Z. Shakir, H. Yanikomeroglu, and A. Yongacoglu, “Backhaul-aware robust 3D drone placement in 5G+ wireless networks,” in Flexible Network Workshop (FlexNet), 2017 IEEE International Conference on Communications, 2017. [10] H. Ahmadi, Y. H. Chew, and C. C. Chai, “A framework to optimize the ofdm-based multiuser dynamic spectrum access networks with frequency reuse,” Wireless personal communications, pp. 1–18, 2013. [11] H. Ahmadi and Y. Chew, “Evolutionary algorithms for orthogonal frequency division multiplexing-based dynamic spectrum access systems,” Computer Networks, vol. 56, no. 14, pp. 3206–3218, 2012. [12] H. Ahmadi and Y. H. Chew, “Adaptive subcarrier-and-bit allocation in multiclass multiuser ofdm systems using genetic algorithm,” in 2009 IEEE 20th International Symposium on Personal, Indoor and Mobile Radio Communications, Sept 2009, pp. 1883–1887. [13] ——, “Subcarrier-and-bit allocation in multiclass multiuser single-cell ofdma systems using an ant colony optimization based evolutionary algorithm,” in 2010 IEEE Wireless Communication and Networking Conference, April 2010, pp. 1–5. [14] R. V. Kulkarni and G. K. Venayagamoorthy, “Particle swarm optimization in wireless-sensor networks: A brief survey,” IEEE Transactions on Systems, Man, and Cybernetics, Part C (Applications and Reviews), vol. 41, no. 2, pp. 262–267, March 2011. [15] A. B. MacKenzie and L. A. DaSilva, “Game theory for wireless engineers,” Synthesis Lectures on Communications, vol. 1, no. 1, pp. 1–86, 2006. [16] L. Gavrilovska, V. Atanasovski, I. Macaluso, and L. A. DaSilva, “Learning and reasoning in cognitive radio networks,” IEEE Communications Surveys Tutorials, vol. 15, no. 4, pp. 1761–1777, Fourth 2013. [17] “Fire project,” https://www.ict-fire.eu.