Gezgin haberleşme sistemleri için yalın zamanlama algoritması

Abstract

Günümüzde gezgin iletişim sistemlerini kullanan kullanıcı sayısı öngörülenin çok ötesinde artmıştır. Üçüncü nesil iletişim sistemlerinin desteklediği yüksek hızlı veri iletimi sayesinde aboneler kablolu sistemler yerine artık gezgin iletişim sistemlerini tercih etmeye başlamışlardır. Kullanıcı sayısındaki yükseliş yanında kullanıcıların tüketmeye başladığı aylık ortalama veri miktarı yüzlerce Mbyte veya GigaByte seviyelerine yükselmiştir. Günümüzde operatörler yüksek veri trafiğini ve hızını karşılamak için 3GPP (3rd Generation Partnership Project) tarafından geliştirilmiş bir standartlar ailesi olan Dördüncü Nesil (4N) gezgin iletişim sistemlerini tercih etmektedirler. 3GPP, bünyesinde ARIB (Association of Radio Industries and Businesses), ATIS (Alliance for Telecommunications Industry Solutions), CCSA (China Communications Standards Association), ETSI (European Telecommunications Standards Institute), TSDSI (Telecommunications Standards Development Society), TTA (Telecommunications Technology Association) ve TTC (Telecommunication Technology Committee) olarak toplam yedi telekom standart geliştirme organizasyonunu birleştiren bir ortaklıktır. İlk ticari 4N sistemi TelliaSoneria tarafından 2009 yılında İsveç'te açılmıştır. GSMA (Global System for Mobile communications Association) birliğinin 2017 yılı istatistiklerine göre gezgin haberleşme şebeklerini kullanan kullanıcı sayısı beş milyarı geçmiştir [1]. Kullanıcı sayısındaki kontrolsüz artış, kullanıcılar arasında etkin bir sistem kaynak paylaşımını daha da önemli kılmaktadır. Yeni nesil akıllı telefonlar ve tablet tipi cihazların yaygınlaşmasıyla kullanıcıların kullanım profili değişmiştir. Kullanıcı profili daha önce uzun süreli az sayıda şebekeye bağlanma ihtiyacı şeklindeyken şimdi az miktardaki veri iletim için daha sık şebekeye bağlanma şekline dönüşmüştür. Kullanıcı profili ve sayındaki bu değişimler halen kullanılmakta olan ve kullanıcıların servis alma paylaşımını düzenleyen zamanlama algoritmalarını verimsiz hale getirmiştir. Yukarda sözü edilen nedenlerle günümüzde üçüncü ve dördüncü nesil gezgin veri iletim sistemlerinde kaynakların paylaştırılmasına yönelik yeni bir zamanlama algoritmasının tasarımına şiddetle ihtiyaç duyulmaktadır. Bu ihtiyaç doğrultusunda bu tezde, sistem kaynaklarının kullanıcılara etkin paylaşımına olanak sağlayacak yeni bir zamanlama algoritmasının geliştirilmesi ve ürün haline getirilmesi hedeflenmiştir. Bu amaçla tezde kullanılacak sistem parametreleri gerçek ve gerçek zamanlı olmayan hizmet alan kullanıcıların ihtiyaçlarına göre belirlenmiş ve kullanıcılara en hızlı/kesintisiz/kaliteli hizmet verebilen dinamik bir zamanlama algoritması geliştirilmiştir.

The number of mobile communication system users have increased far beyond the prescribed at the present days. Subscribers start to prefer mobile communication systems rather than data transmission cable system as a result of high speed data transmission with third-generation mobile system. Besides the increase in the number of subscribers, the average amount of consumed subscriber's data has increased to the level of hundreds of Megabytes or Gigabytes. At the present day, operators prefer the family of standard, Fourth Generation (4G) mobile communication systems which developed by 3GPP (3rd Generation Partnership Project) to support the high data rate traffic and its speed. 3GPP is a partnership that combines seven on-site telecom standards development organizations which are ARIB (Association of Radio Industries and Businesses), ATIS (Alliance for Telecommunications Industry Solutions), CCSA (China Communications Standards Association), ETSI (European Telecommunications Standards Institute), TSDSI (Telecommunications Standards Development Society), TTA (Telecommunications Technology Association) and TTC (Telecommunication Technology Committee). The first commercial 4G system in Sweden was activated in 2009 by TelliaSoneria. According to the statistics of Global System for Mobile communications Association (GSMA), the number of unique subscribers using mobile networks exceeds 5 billion in 2017 [1]. Uncontrolled increase in the number of subscribers makes the effective system resource sharing among users even more important. Subscribers usage profile has changed with the widespread use of the new generation of smart phones and tablet type. These changes in the subscriber profile and number have made the scheduling algorithms which are still being used and regulating the service of subscribers inefficient. For the reasons mentioned above, a new scheduling algorithm in the third and fourth generation mobile data transmission system for the allocation of resources is sorely needed at the moment. In line with the requirements of this thesis, a new scheduling algorithm and its product development are aimed for the effective sharing of system resources. For this purpose, this thesis proposes a new design of the 3GPP LTE (Long Term Evolution) downlink scheduler. Scheduling is one of the essential functions in LTE networks. It interacts intimately and exchanges information with other functionalities such as Link Adaptation and Power Control to use the system resources effectively. In this thesis, a new throughput and CQI (Channel Quality Indicator) aware scheduling algorithm is proposed by integrating Best CQI and PF (Proportional Fair) scheduling algorithms. The integration approach inspired by Lean production method is used for these two scheduling algorithms. The proposed algorithm is called as Lean Scheduler. There is no prior art known to us which applies Lean production concept to LTE downlink scheduling algorithms. Simulation results prove that the proposed new Lean Scheduling algorithm exceeds the fairness performance of Best CQI and the throughput performance of PF algorithms. In addition, the proposed algorithm delivers a sufficient performance to run a communication application at cell edge. Scheduler allows sharing of radio and base station resources between different radio bearers. It provides a trade-off between user fairness and system performance. Considering this expectation, a new delay based Long Term Evolution (LTE) scheduler is also proposed for VoLTE (Voice over LTE) in this thesis. This second proposed algorithm is based on the scheduler having knowledge of the age of Real Time (RT) applications packets. By considering the age of the packet with Lean production concept, the scheduler can determine different type RT applications packets must be scheduled within their packet delay budget. The proposed algorithm is called as Delay Based Lean Scheduler. Simulation results shows that the proposed Delay Based Lean Scheduling algorithm provides a superior throughput and delay performance than M-LWDF (Maximum-Largest Weighted Delay First ) algorithm independent from the number of user. M-LWDF is the most common used delay based algorithm in literature. The term 'lean' is popularized in Europe, which clearly explains the significant performance gap between Toyota Production System (TPS) and western automotive production systems. It defines the key elements accounting for this superior performance as lean production. The core idea of lean aims maximizing customer value while minimizing waste. Lean means creating more value for customers with fewer resources.Idea of lean can be applied to any industry or sector even though lean is generated for the automotive manufacturing environment. Lean efficiency matrix is the key factor to understand the Lean philosophy. Our Lean scheduler proposal to LTE resource allocation problem builds on resource and flow efficiencies in Lean matrix. Radio resource usage efficiency and fairness in LTE are mapped to resource and flow efficiencies in Lean matrix respectively. Our Lean scheduler strategy obtains a balance between flow efficiency and resource efficiency effectively.The downlink scheduling in LTE allocates UEs to Resource Blocks (RBs) using the input parameters such as highest signal to noise ratio, order of the modulation, coding and spatial multiplexing. LTE scheduler is responsible for the distribution of control and user data on the physical resource blocks in time and frequency domains across the radio interface. It multiplexes and schedules the subscribers simultaneously for efficient use of radio resources. A scheduler is able to give a sufficient performance to run a communication application at an acceptable level of quality. A bounded scalar metric which is generated for kth user in response to a real-time measurement for CQI, instantaneous and average throughputs is defined. This metric is used as ranking operation in the proposed scheduler. It is then optimized to guarantee a sufficient performance to run a communication application at an acceptable cell edge quality. After obtaining metric value for all users, the proposed scheduler starts allocating radio resources in a rank order methodology during the total simulation time in Transmission Time Intervals (TTIs). The second proposed algorithm for RT traffic determines that the real time conversational and media consumption packets must be prioritized over non-real time user while non-real time users have minimum acceptable throughput. Non-real time users are social and sharing application ones. The proposed scheduler results in an efficient scheduling of users. Setting a high value for packet delay budget may result in a reduction of the real time conversational application quality experienced by the end users. On the other hand, setting a small value for packet delay budget may result in a reduced average uplink and downlink throughput for non-real time users. The priority queue weights define the prioritization of the users according to age of data, packet delay, quality and throughputs. There are three priority queue weights for proposed delay based scheduling. Delay based weight is calculated according to the age of the data and the packet delay budget. The older data gets higher weight. Channel quality weight is calculated based on the UE reported channel quality. Higher channel quality gives higher weight. The last weight is calculated from the instantaneous and average throughput values. The corresponding priority queue weights shall be calculated according to the requirements set in the proposed scheduling algorithm. The weight of RT and non-RT type applications changes differently. The delay based scheduled users will not have priority over non-real time users until it is necessary. After obtaining metric value for all RT and non-RT users, the proposed scheduler starts allocating radio resources in a rank order methodology.