Business Understanding
In the dynamic realm of 5G, network operators are presented with a unique opportunity to dynamically craft and deploy a diverse range of services, such as massive IoT, mMTC, and eMBB. Yet, this heightened flexibility and complexity necessitate solutions that uphold security, reliability, and resource allocation in a dynamically robust and dependable manner. Here, the pivotal role of AI technology comes into play, serving as the cornerstone that empowers the future networks, encompassing 5G and beyond.
Three distinct domains emerge as the arena where AI technology reshapes the landscape: Network Planning, Network Diagnostics and Insights, and Network Optimization and Control. Two pivotal challenges underscore this transformation. Firstly, the endeavor of placing probes at every network point for data collection becomes impractical. Secondly, the massive and sparse operational data, a treasure trove of insights, poses a formidable challenge for human experts to digest effectively, given the limitations of traditional theoretical and empirical processing.
Amidst the surging demands of transportation, smart cameras, media streaming, and beyond, the accurate prediction of data traffic volume emerges as a linchpin for Service Providers to honor their Service Level Agreements. Yet, the intricacies of dense urban environments and the ever-shifting landscape of mobile traffic due to user mobility render the placement of probes complex and expensive. In this intricate dance of variables, it becomes evident that AI technology stands as the sole reliance for 5G and future Networks, supplanting human expertise and steering the course towards transformative possibilities.
AI Solution
- To surmount these complexities, a promising avenue lies in the adoption of a Deep Spatio-Temporal Network (D-STN).
- The architecture involves leveraging 3D convolution layers to discern the intrinsic learnable features inherent in spatio-temporal traffic data. This process effectively unravels the underlying patterns that govern the movement of data across space and time.
- Further enhancing the predictive prowess, Convolution Long Short Term Memory (ConvLSTM) units come into play. By integrating the strengths of both convolutional and LSTM structures, these units facilitate accurate and dependable long-term projections of mobile traffic patterns.
- To round off the model’s capabilities, Fully Connected Layers are employed. These layers function as the final touchpoint, effectively regressing the projected traffic volume. In doing so, they translate the complex web of learnings and predictions into a comprehensive and actionable forecast.
Traffic volume in one cell forecast over 10 hours (after 2-hour observations) with Holt-Winters exponential smoothing, ARIMA, and D-STN, which captures spatio-temporal correlations .
Mobile Traffic Forecasting for one cell
Expected Outcome
In the realm of Mobile Traffic Forecasting, the fusion of AI technology with the intricate dynamics of data holds transformative potential. The utilization of a Deep Spatio-Temporal Network (D-STN) stands as a beacon of innovation, effectively addressing the complex challenges that mobile traffic prediction poses. With the strategic deployment of 3D convolution layers, Convolution Long Short Term Memory (ConvLSTM) units, and Fully Connected Layers, the AI-driven framework becomes a powerful predictive tool. It discerns hidden patterns within spatio-temporal traffic data, crafts reliable long-term projections, and ultimately regresses accurate traffic volume predictions. Through this amalgamation of cutting-edge AI methodologies, Mobile Traffic Forecasting transcends conventional limitations, offering a sophisticated and adaptable solution that paves the way for smarter, data-driven decision-making in the fast-evolving landscape of mobile networks.
Our Team Members
AI TRANSFORMATION LEAD
Dr. Mikhail Krasnov
Based in Switzerland, Mikhail is an esteemed leader with a Ph.D. in Economics and recognized as a Certified Director by the London Institute of Directors. With a 30-year tenure in the IT sector, he co-founded Verysell Group in 1990 and now drives strategy and innovation across the entire organization. He staunchly believes that AI should be a catalyst for all businesses, not a threat, enhancing and transforming companies. To him, AI is a powerful tool for businesses to harness for their advantage.
Head of Applied AI Lab
Yurii Lozinskyi
Based in England, Yurii is the driving force behind Verysell Applied AI Lab. With 25 years of experience in Healthcare, Insurance, and Finance, he specializes in Business Process & Digital Transformation. Yurii excels in creating neural networks for diverse business automation, from medical diagnostics to machinery anomaly detection. Beyond tech, he’s a proficient software delivery manager and project leader. His leadership has fueled remarkable growth, achieving up to 60% CAGR by guiding 20+ Agile Teams globally.
Chief AI Scientist
Dr. Hung Dao
Based in Vietnam, Dr. Hung leads a dynamic Consulting Team, driving AI innovation across diverse sectors, from industrial visual inspection to medical image segmentation at our AAI Lab. With a PhD in Engineering from Keio University (2014), his expertise extends to groundbreaking research published in renowned conferences such as FG, BMVC, ICASSP, and ACCV, alongside multiple granted patents. With a distinguished track record at FPT, VinBrain & Nautilus AI, he’s a prominent figure in the AI landscape.
STRATEGIC ADVISOR ON AI
Dr. Alexey Minin
Prof. Dr. Alexey Minin, an esteemed expert in the digital economy and artificial intelligence implementation, boasts over 19 years in high-tech. Rising from Siemens engineer to associated partner at MHP – A Porsche Company, Germany, Alexey excels in management consulting, exponential technologies, and creating businesses based on modern tech. A Doctor of Science in AI, Honored Professor, and consultant for several boards of directors, Alexey specializes in digital transformation, AI application, and startup guidance. He co-founded startups, lectures globally, and is a sought-after speaker in digital conferences.
Business Development Manager
Muin Pirzada
Muin brings to the table an extensive background of over 13 years in the fields of Robotic Process Automation (RPA), Test Automation, and Artificial Intelligence (AI). This vast experience has not only equipped him with a profound knowledge of these cutting-edge technologies and deep passion for business growth and operational efficiency, but also with an acute understanding of their capability to revolutionize industries and streamline operations.
Marketing Manager
Ha Nguyen
Based in Hanoi, Ha spearheads marketing endeavors, leveraging over a decade of prowess in the global technology arena. She currently serves as AAI Lab’s Marketing Manager, leveraging her expertise in B2B marketing, branding, content strategies, and Agile methodologies to propel the company’s growth. Her practical know-how equips her to craft and execute effective marketing strategies for AAI Lab and other entities within the Verysell Group.
Software Development Automation Lead
Hieu Tran
Hieu, from Danang, Vietnam, heads the automation team at SmartDev. With a decade of experience in software development, he’s skilled in tech strategy and research. Known for boosting daily operations, Hieu shines in project coordination and customer happiness. Currently, he’s the top engineer at SmartDev and also leads the AAI Lab Software Development Automation.
Head of Project Management
Richard Sharp
Richard, based in Danang Vietnam, has over 40 years of IT experience. He has had many roles including Client Manager, Supplier Manager, and Service Management Program Manager. He has designed, developed and implemented advanced service management solutions for leading organizations across the technology, government, health, telecoms, oil & gas and financial services sectors.
HEAD OF AI EDUCATION
Soan Duong
Dr. Soan is a lecturer at the Computer Science Department within Le Quy Don Technical University. She is also a senior research scientist with keen interest in computer vision and medical imaging. Her research work has been published at prestigious venues, including ISBI, ICIP, ICASSP, BMVC, and CVPR. Soan plays an active role in the academic community by serving as a dedicated reviewer for esteemed publications such as: IEEE Access and NMR in Biomedicine.