:1. the complete title of one (or more) paper(s) published in the open literature describing the work that the author claims describes a human-competitive result; Automated Self-Optimization in Heterogeneous Wireless Communications Networks ---------------------------------- ---------------------------------- 2. the name, complete physical mailing address, e-mail address, and phone number of EACH author of EACH paper(s); David Lynch, UCD NCRA, Building D, UCD Michael Smurfit Graduate School of Business, Carysfort Avenue, Blackrock, Co. Dublin, Ireland. david.lynch@ucd.ie +353 87 313 5346 Michael Fenton, Corvil, Dublin, D01 K5C7 Ireland, Co. Dublin, Ireland. michael.fenton@ucd.ie +353 86 326 5665 David Fagan, UCD NCRA, Building D, UCD Michael Smurfit Graduate School of Business, Carysfort Avenue, Blackrock, Co. Dublin, Ireland. david.fagan@ucd.ie +353 86 881 6078 Stepan Kucera Nokia Bell Labs at Dublin, Co. Dublin, D15 Y6NT, Ireland stepan.kucera@nokia-bell-labs.com Holger Claussen Nokia Bell Labs at Dublin, Co. Dublin, D15 Y6NT, Ireland holger.claussen@nokia-bell-labs.com +353 87 661 5789 Michael O'Neill, UCD NCRA, Building D, UCD Michael Smurfit Graduate School of Business, Carysfort Avenue, Blackrock, Co. Dublin, Ireland. m.oneill@ucd.ie +353 86 859 2986 ---------------------------------- ---------------------------------- 3. the name of the corresponding author (i.e., the author to whom notices will be sent concerning the competition); David Lynch ---------------------------------- ---------------------------------- 4. the abstract of the paper(s); Traditional single-tiered wireless communications networks cannot scale to satisfy exponentially rising demand. Operators are increasing capacity by densifying their existing macro cell deployments with co-channel small cells. However, cross-tier interference and load balancing issues present new optimisation challenges in channel sharing heterogeneous networks (HetNets). One-size-fits-all heuristics for allocating resources are highly suboptimal, but designing ad hoc controllers requires significant human expertise and manual fine-tuning. In this paper, a unified, flexible, and fully automated approach for end-to-end optimisation in multi-layer HetNets is presented. A hill climbing algorithm is developed for reconfiguring cells in real time in order to track dynamic traffic patterns. Schedulers for allocating spectrum to user equipment are automatically synthesised using grammar-based genetic programming. The proposed methods for configuring the HetNet and scheduling in the time–frequency domain can address ad hoc objective functions. Thus, the operator can flexibly tune the tradeoff between peak rates and fairness. Far cell edge downlink rates are increased by up to 250% compared with non-adaptive baselines. Alternatively, peak rates are increased by up to 340%. The experiments illustrate the utility and future potential of natural computing techniques in software-defined wireless communications networks. ---------------------------------- ---------------------------------- 5. a list containing one or more of the eight letters (A, B, C, D, E, F, G, or H) that correspond to the criteria (see above) that the author claims that the work satisfies; (B) The result is equal to or better than a result that was accepted as a new scientific result at the time when it was published in a peer-reviewed scientific journal. (D) The result is publishable in its own right as a new scientific result — independent of the fact that the result was mechanically created. (E) The result is equal to or better than the most recent human-created solution to a long-standing problem for which there has been a succession of increasingly better human-created solutions. ---------------------------------- ---------------------------------- 6. a statement stating why the result satisfies the criteria that the contestant claims (see examples of statements of human-competitiveness as a guide to aid in constructing this part of the submission); (B) The result is equal to or better than a result that was accepted as a new scientific result at the time when it was published in a peer-reviewed scientific journal. HetNet operators such as Vodafone and AT&T attract and retain customers (hereafter "users") by providing high downlink rates (i.e. high throughput). It is especially important to increase throughput at cell edges, where interference from nearby cells can be prohibitive. Cell edge throughput can be managed through a combination of (1) load balancing, (2) interference mitigation, and (3) scheduling. These mechanisms are implemented in the 4G LTE standard as follows: (1) Load Balancing: the number of users (i.e. load) on the Macro Cell and Small Cell tiers is balanced by modulating Small Cell powers and biases. (2) Interference mitigation: severe interference at Small Cell edges is managed by periodically muting high-powered Macro Cells. (3) Scheduling: cells compute new schedules every 40 milliseconds. A schedule specifies how the limited spectrum will be allocated to all users that are served by a cell. Thus, a HetNet has several parameter 'layers' which must be jointly optimised. Layers 1-3 refer to the HetNet 'configuration', that is, the Small Cell biases and powers, and the Macro Cell muting patterns respectively. A HetNet is reconfigured every few minutes as traffic patterns change. Layer 4 refers to the schedules that are executed by Macro Cells and Small Cells. Our paper presents better than human competitive algorithms for optimising multi-layer HetNets. A hill climbing algorithm is proposed for load balancing and interference mitigation. Our algorithm achieves significantly higher cell-edge throughput than the best self-organising network (SON) algorithms from the literature [1]. Schedulers that we evolve using Grammar-based Genetic Programming (GP) outperform a state of the art manually engineered scheduler [2]. Human-designed algorithms address sub-problems of the overall optimisation task. For instance, the authors in [17] tackle load balancing only. Separate algorithms are presented for load balancing, interference mitigation, and scheduling in [2]. However, specialised controllers tend to produce contradictory control actions when combined. Our main contribution is a fully unified approach that generates compatible settings across all layers of a HetNet. Our data-driven algorithms produce tailored strategies. For example, schedulers evolved using GP outperform a state of the art benchmark [2] because they are tailored to the deployment scenario. Automation mitigates the need for costly manual fine-tuning of inflexible one-size-fits-all algorithms. Finally, service differentiation is a core feature of the LTE standard and the ongoing 5G standardisation. The authors in [3] state that their algorithms could support service differentiation, but they do not provide an implementation. We instrument a novel fitness function that enables precise control over fairness tradeoffs. Hence, an operator can decide how aggressively peak throughput should be sacrificed for the sake of cell edge throughput. Far cell edge throughput is increased by up to 250% compared with non-adaptive baselines. Alternatively, peak rates can be increased by up to 340%. In summary, the techniques we present in our paper outperform the best human-designed algorithms. Moreover, they constitute a fully automated and unified approach to self-optimisation in multi-layer HetNets. Our flexible framework for configuring cells and scheduling replaces costly and inefficient design by human experts. ---------------------------------- (D) The result is publishable in its own right as a new scientific result — independent of the fact that the result was mechanically created. Our result was published in the IEEE/ACM Transactions on Networking journal [14]. This is the leading journal in the application domain that is "devoted to the timely release of high quality papers that advance the state-of-the-art of [wireless communications] networks". Thus, accepted papers must meet high standards of scientific rigour, and they typically describe a novel contribution that outperforms existing technology. The approach and results were well received by the reviewers. One reviewer stated that "The joint optimisation of small cell configuration and scheduling is complex, and the proposed optimisation solution is of interest." Another reviewer remarked that "The paper is well written with a very clear logic and tons of practical results." Our novel application of GP to address the scheduling problem in HetNets has also given rise to a patent application (International Publication Number WO 2018/149898 A2, International Application Number PCT/EP2018/053730, 15 February 2018). This patent filing confirms that the work is of significant commercial value. Furthermore, the research building up to our publication in the IEEE/ACM Transactions on Networking journal has been published in numerous conferences [4][5][6][7][8][9][10][11] and journals [12][13] dedicated to evolutionary computation. Our EuroGP paper [10] was nominated for a best paper award. These publications suggest that the work is of interest to the wider operations research community. ---------------------------------- (E) The result is equal to or better than the most recent human-created solution to a long-standing problem for which there has been a succession of increasingly better human-created solutions. The cellular concept was invented by Bell Laboratories in the 1950s. The first generation (1G) analogue cellular networks emerged during the 1970s. Faster and more secure digital 2G systems were rolled out in the 1980s. The 3G networks of the late 1990s made internet browsing possible on mobile devices. These early standards introduced ever more innovative ways of utilising perennially scarce wireless spectrum. However, exploding traffic from smartphones during the mid 2000s pushed 3G networks to their breaking point. The HetNet concept was proposed in the 4G LTE standard as a solution to the problem of exponentially growing traffic. Therein, the mechanisms of load balancing, interference mitigation, and scheduling were initially examined at a conceptual level [15][16]. However, these pilot studies did not specify any algorithmic implementations. Algorithms were subsequently developed for load balancing in HetNets [17]. Next, joint algorithms for load balancing and interference mitigation were proposed [3][19]. The scheduling problem was concurrently addressed using greedy strategies [2], dynamic programming [20], game theory [21]. Our more unified approach jointly optimises all layers of a HetNet. Furthermore, the cited works focus on maximising cell edge throughput only, whereas our algorithms strike arbitrary fairness tradeoffs. Hence, an operator can devise bespoke marketing campaigns (e.g. advertising peak throughput versus network-wide reliability). ---------------------------------- ---------------------------------- 7. a full citation of the paper (that is, author names; publication date; name of journal, conference, technical report, thesis, book, or book chapter; name of editors, if applicable, of the journal or edited book; publisher name; publisher city; page numbers, if applicable); Lynch, David; Fenton, Michael; Fagan, David; Kucera, Stepan; Claussen, Holger; O'Neill, Michael, "Automated Self-Optimization in Heterogeneous Wireless Communications Networks." IEEE/ACM Transactions on Networking 27.1 (2019): 419-432. @article{lynch2019automated, title={Automated Self-Optimization in Heterogeneous Wireless Communications Networks}, author={Lynch, David and Fenton, Michael and Fagan, David and Kucera, Stepan and Claussen, Holger and O’Neill, Michael}, journal={IEEE/ACM Transactions on Networking}, volume={27}, number={1}, pages={419--432}, year={2019}, publisher={IEEE} } ---------------------------------- ---------------------------------- 8. a statement either that "any prize money, if any, is to be divided equally among the co-authors" OR a specific percentage breakdown as to how the prize money, if any, is to be divided among the co-authors; The prize money, if any, is to be awarded to the lead author who will disseminate it amongst the co-authors. ---------------------------------- ---------------------------------- 9. a statement stating why the authors expect that their entry would be the "best": We are confident that this entry qualifies as a winning entry for a number of reasons: - The work has been published in leading journals in both the field of evolutionary computation [12][13] and in the application domain [14]. It has also resulted in a patent application. - Our entry addresses the core problem at the heart of modern 4G and 5G wireless communications networks. There has been a recent paradigm shift away from ad hoc manually engineered solutions towards data-driven automated self-optimisation. Our paper illustrates the suitability of evolutionary algorithms for managing software-defined networks. - The novel application of evolutionary algorithms can make a major impact in the multi-trillion dollar telecommunications industry. Realising the 5G vision of ultra-fast wireless communications will improve the quality of life of billions of people. Nascent technologies like virtual reality and self-driving cars will require reliable networks with near zero latency, which will depend upon zero-touch self-optimisation approaches, such as our approach published in [14]. - We have found that evolutionary algorithms outperform the best human-designed benchmarks. However more significantly, our work introduces a new paradigm for managing wireless communications networks based on evolutionary computation. Evolutionary algorithms lend themselves well to this task for the following reasons: (1) Firstly, techniques like GP can automatically construct high-performance and tailored algorithms for network control given only minimal domain knowledge. Automation will become increasingly important as wireless deployments become more complex, decentralised, and heterogenous. For instance, complex non-linear interactions, highly dynamic environments, and a variety of interacting technologies already render manual design infeasible in some 5G deployments. (2) Secondly, ad hoc fitness functions can be designed that reflect an operator's business objectives. For instance, we demonstrated how fairness tradeoffs could be managed in our paper [14]. (3) Thirdly, evolutionary algorithms cope well in the dynamic and uncertain environments that characterise a wireless network. ---------------------------------- ---------------------------------- 10. An indication of the general type of genetic or evolutionary computation used, such as GA (genetic algorithms), GP (genetic programming), ES (evolution strategies), EP (evolutionary programming), LCS (learning classifier systems), GE (grammatical evolution), GEP (gene expression programming), DE (differential evolution), etc: Grammar-based Genetic Programming ---------------------------------- ---------------------------------- 11. the date of publication of each paper. If the date of publication is not on or before the deadline for submission, but instead, the paper has been unconditionally accepted for publication and is "in press" by the deadline for this competition, the entry must include a copy of the documentation establishing that the paper meets the "in press" requirement. February 2019 ---------------------------------- ---------------------------------- References: [1] Tall, Abdoulaye, Zwi Altman, and Eitan Altman. "Self organizing strategies for enhanced ICIC (eICIC)." 2014 12th International Symposium on Modeling and Optimization in Mobile, Ad Hoc, and Wireless Networks (WiOpt). IEEE, 2014. [2] López-Pérez, David, and Holger Claussen. "Duty cycles and load balancing in HetNets with eICIC almost blank subframes." 2013 IEEE 24th International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC Workshops). IEEE, 2013. [3] Deb, Supratim, et al. "Algorithms for enhanced inter-cell interference coordination (eICIC) in LTE HetNets." IEEE/ACM Transactions on Networking (ToN) 22.1 (2014): 137-150. [4] Lynch, David; Fagan, David; Kucera, Stepan; Claussen, Holger; O'Neill, Michael, “Managing Quality of Service through Intelligent Scheduling in Heterogeneous Wireless Communications Networks.” IEEE World Congress on Computational Intelligence. 2017. [5] Fenton, Michael; Lynch, David; Kucera, Stepan; Claussen, Holger; O'Neill, Michael, “Load balancing in heterogeneous networks using an evolutionary algorithm." Evolutionary Computation (CEC), 2015 IEEE Congress on. IEEE, 2015. [6] Fagan, David; Fenton, Michael; Lynch, David; Kucera, Stepan; Claussen, Holger; O'Neill, Michael, "Deep learning through evolution: A hybrid approach to scheduling in a dynamic environment." Neural Networks (IJCNN), 2017 International Joint Conference on. IEEE, 2017. [7] Lynch, David; Fenton, Michael; Kucera, Stepan; Claussen, Holger; O'Neill, Michael, "Configuring Dynamic Heterogeneous Wireless Communications Networks Using a Customised Genetic Algorithm." European Conference on the Applications of Evolutionary Computation. Springer, Cham, 2017. [8] Lynch, David; Fenton, Michael; Kucera, Stepan; Claussen, Holger; O'Neill, Michael, “Ensemble Techniques for Scheduling in Heterogeneous Wireless Communications Networks,” in Operations Research Proceedings. Springer, 2016, in press. [9] Lynch, David; Fenton, Michael; Kucera, Stepan; Claussen, Holger; O'Neill, Michael, "Evolutionary Learning of Scheduling Heuristics for Heterogeneous Wireless Communications Networks." Proceedings of the 2016 Conference on Genetic and Evolutionary Computation Conference. ACM, 2016. [10] Lynch, David; Fenton, Michael; Kucera, Stepan; Claussen, Holger; O'Neill, Michael, "Scheduling in Heterogeneous Networks Using Grammar-Based Genetic Programming." European Conference on Genetic Programming. Springer, Cham, 2016. [11] Fenton, Michael; Lynch, David; Kucera, Stepan; Claussen, Holger; O'Neill, Michael, "Evolving coverage optimisation functions for heterogeneous networks using grammatical genetic programming." European Conference on the Applications of Evolutionary Computation. Springer, Cham, 2016. [12] Fenton, Michael; Lynch, David; Kucera, Stepan; Claussen, Holger; O'Neill, Michael. "Multilayer Optimization of Heterogeneous Networks Using Grammatical Genetic Programming." IEEE transactions on cybernetics 47.9 (2017): 2938-2950. Conference Procedings: [13] Fenton, Michael; Lynch, David; Fagan, David; Kucera, Stepan; Claussen, Holger; O'Neill, Michael. "Towards Automation & Augmentation of the Design of Schedulers for Cellular Communications Networks." Evolutionary computation Just Accepted (2018): 1-30. [14] Lynch, David; Fenton, Michael; Fagan, David; Kucera, Stepan; Claussen, Holger; O'Neill, Michael, "Automated Self-Optimization in Heterogeneous Wireless Communications Networks." IEEE/ACM Transactions on Networking 27.1 (2019): 419-432. [15] David Lopez-Perez, Ismail Guvenc, Guillaume De la Roche, Marios Kountouris, Tony QS Quek, and Jie Zhang. Enhanced intercell interference coordination challenges in heterogeneous networks. IEEE Wireless Communications, 18(3):22–30, 2011. [16] Kenta Okino, Taku Nakayama, Chiharu Yamazaki, Hirotaka Sato, and Yoshimasa Kusano. Pico cell range expansion with interference mitigation toward LTEAdvanced heterogeneous networks. In Communications Workshops (ICC), 2011 IEEE International Conference on, pages 1–5. IEEE, 2011. [17] Steven Corroy, Laetitia Falconetti, and Rudolf Mathar. Dynamic cell association for downlink sum rate maximization in multi-cell heterogeneous networks. In Communications (ICC), 2012 IEEE International Conference on, pages 2457–2461. IEEE, 2012. [18] David Lopez-Perez, Xiaoli Chu, and ˙Ismail Guvenc. On the expanded region of picocells in heterogeneous networks. IEEE Journal of Selected Topics in Signal Processing, 6(3):281–294, 2012. [19] Anand Bedekar and Rajeev Agrawal. Optimal muting and load balancing for eICIC. In Modeling & Optimization in Mobile, Ad Hoc & Wireless Networks (WiOpt), 2013 11th International Symposium on, pages 280–287. IEEE, 2013 [20] Jiyong Pang, Jun Wang, Dongyao Wang, Gang Shen, Qi Jiang, and Jianguo Liu. Optimized time-domain resource partitioning for enhanced inter-cell interference coordination in heterogeneous networks. In Wireless Communications and Networking Conference (IEEE WCNC), pages 1613–1617. IEEE, 2012. [21] Lei Jiang and Ming Lei. Resource allocation for eICIC scheme in heterogeneous networks. In 23rd International Symposium on Personal Indoor and Mobile Radio Communications (IEEE PIMRC), pages 448–453. IEEE, 2012.