In the last years, the information and communication technology (ICT) sector has transformed the way we live. Consistently delivering innovative products and services, the ICT assumed a primary role on economic development and productivity, becoming an integral part of everyday life. However, due to their wide and constantly increasing diffusion, the effect of information and communication technologies on global warming can no longer be ignored. The concept of Green ICT has originated with the aim of building awareness of this, thus boosting the research toward environmentally sustainable, energy-efficient technologies and solutions. As an important part of the ICT, telecommunication networks are experiencing a booming growth. Capacity issues and quality of service constraints are some of the main concerns that contribute to raise the power consumption. In particular, a large portion of the electricity bill results from the high power requirements of wireless base stations, which have been proved to be the most energy-hungry network components. Up to now, the mobile communication industry has focused mostly on the development of power-efficient mobile terminals, so as to attract a higher number of customers and con- sequently increase the operators’ profits; on the other hand, the research world has been investigating energy efficiency from a wider point of view. Besides studies on power-efficient devices and protocols, more recent works addressed the problem of energy-aware design and operation in wired and wireless network infrastructures. Many aspects and challenges of green network planning and management have been explored; nevertheless, the two problems have never been linked and tackled at the same time, neglecting the fact that an effective power-efficient network operation largely depends on the decisions taken in the design phase. The research presented in this doctoral thesis aims at filling this gap by developing an optimization framework that jointly considers the design and operation of wireless networks. The proposed joint planning and energy management problem (JPEM) strives to prove that, when cell sleeping is adopted as network management technique, the level of flexibility offered by the installed topology strongly improves the system capability to adapt to the varying traffic load. By minimizing the trade-off between capital expenditures (CapEx) related to the network deployment and operational and management expenditures (OpEx) calculated over the network lifetime, the model finds the most energy-efficient network topology while meeting the capital investment limitations imposed by the mobile operator. Two types of access network are analyzed in this work. As the name suggests, the joint planning and energy management problem for cellular networks (JPEM-CN) was developed to plan and manage the operation of cellular networks. Three cell sizes (macro, micro and pico) are allowed to be deployed in an area where no previous access devices are installed. A realistic daily traffic profile is assumed to characterize the area; therefore, the traffic offered by the test points randomly distributed in the area varies in each time period. Numerical results obtained by testing six different scenarios demonstrate that, with modest CapEx increases, a network planned to be energy efficient can reach power savings around 50%–60% compared to the energy saved by managing the operation of the minimum cost deployment. Moreover, the results give an interesting insight on the different topology compositions. While the minimum CapEx network is mostly composed of big and medium size cells, energy-aware topologies include a few larger cells in support of many small cells, which can be put to sleep during low traffic periods without leaving parts of the area uncovered. Besides JPEM-CN, another optimization framework has been modeled to solve the joint design and operation problem on wireless mesh networks (JPEM-WMN). In this case, only two types of access station can be placed in the area: routers, that can only connect to other routers, and gateways, having direct access to the Internet. Once again, a realistic traffic profile is considered, according to which the mesh clients vary their traffic requests; also, two degrees of traffic congestion are examined, standard and busy. The joint framework were tested on six scenarios. Even though the same benefits of the JPEM-CN apply to mesh networks, smaller energy savings are registered due to the lower flexibility of WMNs in comparison to heterogeneous cellular networks. However, the JPEM-WMN model yields savings of 25% to 30% in most cases with low extra installation costs. Additional model variants for both JPEM models are also presented in this research. Since one of the objective of the work is to evaluate the impact of using a joint modeling framework compared to a more traditional CapEx optimization and successive network management, a two-step approach is developed where the minimum cost network is first installed and then operated. The energy savings served to fairly evaluate the reduction of power consumption achieved by the cellular and mesh versions of JPEM. Another important variation, formulated for both types of access technology, consists in relaxing the total area coverage constraints to guarantee network service only to the active test points or mesh clients. This way, access devices having only idle users in their coverage radius can be put to sleep, further decreasing the energy consumption: results showed that supplemental power savings of 25%–35% can be reached in cellular test scenarios, while the percentage is around 12% in case of mesh instances. Referring now only to cellular networks, this thesis includes a JPEM-CN variant where the framework is not free to select the best topology (and thus, capital investments) according to the relative importance of the energy saving component in the network planning. Instead, a budget parameter is imposed in order to limit the CapEx to a certain value. The Budget JPEM-CN achieves similar results to the original version; however, the new constraints capping the capital costs seem to add complexity to the model formulation. For what concerns mesh networks, an interesting On/Off JPEM-WMN framework is presented where the access devices can switch their state, from idle to active and vice versa, only once per day, reducing the power wasted during state transitions but inevitably decreasing the energy savings from sleeping cells. Further minor changes have been also tested on the JPEM-WMN during the course of the doctoral research, as the introduction of variable backbone link capacity, compared to the fixed capacity in the original model, and the elimination of the multi-hop connectivity characteristics in order to simulate the behavior of a cellular network, by allowing only the installation of mesh gateways. In order to simplify the solution of the JPEM-CN and enable the evaluation of test scenarios closer to real size, an ad-hoc heuristic method was developed where the planning and operation problems, as well as the daily time intervals, are tackled separately. Starting from a complete topology, the heuristic finds the most efficient activation pattern that satisfies the coverage requirements for both the peak traffic periods, where high traffic volumes have to be served, and the off-peak, where the opportunity to save energy by turning off some access stations is the highest. The resulting activation pattern is considered as a new partial topology, which is provided as input to the original JPEM-CN; the initial topology is then enriched to obtain the best feasible solution for all time periods. For the same cellular test scenarios, the heuristic showed results only about 2% to 13% far from the respective JPEM-CN objective functions; on the other hand, new larger scenarios have been tested and successfully solved in less than 15 minutes in most of the cases. On the whole, the various JPEM formulations proved that higher energy savings can be obtained when the network topology is designed to be power efficient at the cost of moderate increases in capital costs. To do so, the effects of the network management have to be taken into consideration during the network design stages. Moreover, numerical results and examples showed how the coexistence of multiple sizes of access device in the same topology is fundamental to provide the network with enough flexibility to adapt to the traffic variations in time and space.
Nel ultimi anni, grazie alla continua offerta di prodotti e servizi innovativi divenuti parte integrante della nostra quotidianità, il settore delle tecnologie dell’informazione e delle comunicazioni (information and communication technology, ICT) ha assunto un ruolo fondamentale nello sviluppo economico. La continua crescita e diffusione delle ICT non ha solo trasformato il nostro stile di vita, ma anche attirato l’attenzione sull’impatto delle tecnologie dell’informazione e comunicazione sul problema del surriscaldamento globale. In questo contesto di accresciuta consapevolezza è nato il concetto di Green ICT, che ha come scopo la ricerca di tecnologie e soluzioni eco-sostenibili ed energeticamente efficienti. Come parte fon- damentale delle ICT, anche le reti di telecomunicazioni sperimentano una costante crescita ed espansione. I vincoli di capacità e qualità del servizio offerto sono tra i maggiori responsabili dell’aumento del consumo di energia delle reti. Per quanto riguarda le reti senza fili in particolare, gran parte delle spese sostenute dagli operatori è dovuta alle alte esigenze energetiche delle stazioni di base, considerate tra le componenti di rete con il più alto assorbimento di potenza. Fino ad oggi, l’industria delle comunicazioni senza fili si è concentrata essenzialmente sulla produzione di terminali mobili a basso consumo energetico al fine di attirare un maggior numero di clienti e, di conseguenza, aumentare i profitti degli operatori di rete. D’altra parte, negli ambienti di ricerca, la questione dell’efficienza energetica è affrontata da un punto di vista più ampio. Oltre allo sviluppo di dispositivi e protocolli ad alta efficienza energetica, studi più recenti riguardano la pianificazione e l’operazione energy-aware di reti wireless. I problemi di design e gestione di reti green sono stati trattati sotto molteplici aspetti, ma mai in maniera congiunta, trascurando il fatto che l’efficacia di un’operazione di rete a basso consumo energetico dipende strettamente dalle decisioni prese in fase di pianificazione. Il lavoro di ricerca presentato in questa tesi di dottorato mira a colmare questa lacuna tramite lo sviluppo di un sistema di ottimizzazione che considera unitamente il design e l’operazione delle reti senza fili. Il problema congiunto di pianificazione e gestione energetica (joint planning and energy management problem, JPEM) proposto punta a dimostrare che, quando il meccanismo di cell sleeping è utilizzato come tecnica di gestione di rete, il livello di flessibilità offerto dalla topologia installata aumenta fortemente la capacità della rete di adattarsi alle variazioni del traffico offerto. Minimizzando il trade-off tra gli investimenti iniziali (capital expenditures, CapEx) dovuti all’installazione delle stazioni di base, e le spese di operazione e gestione (operational and management expenditures, OpEx), calcolate sulla vita della rete, il modello stima la topologia a più alta efficienza energetica ottenibile rispettando eventuali vincoli imposti dall’operatore di rete sulle spese di capitale. In questo studio sono prese in analisi due tecnologie di rete d’accesso. Come intuibile dal nome, il problema congiunto di pianificazione e gestione energetica per reti cellulari (joint planning and energy management problem for cellular networks, JPEM-CN) è stato sviluppato per il design e la gestione dell’operazione di reti cellulari. Tre tipi di celle (macro, micro e pico) possono essere utilizzate per la copertura di un’area dove non vi è alcuna stazione d’accesso preesistente. La zona in esame è caratterizzata da un profilo di traffico giornaliero realistico; di conseguenza, il traffico offerto dai punti di concentrazione (test points, TPs) distribuiti in maniera aleatoria nell’area viaria a seconda del periodo di tempo considerato. I risultati numerici ottenuti testando sei scenari differenti mostrano che, al costo di un leggero aumento in CapEx, una rete pianificata per essere energicamente efficiente permette di risparmiare circa il 50%–60% rispetto all’energia necessaria all’operazione della topologia di costo minimo. Dai risultati appare inoltre che, mentre le reti a costo minimo sono composte prevalentemente da celle di grandi e medie dimensioni, le topologie ad alta efficienza energetica comprendono alcune macro celle a supporto di numerose micro e pico celle, le quali possono essere spente facilmente durante i periodi di minor traffico senza violare i vincoli di copertura totale. Un altro sistema di ottimizzazione è stato modellizzato per risolvere il problema congiunto di pianificazione e gestione energetica per reti a maglie senza fili (joint planning and energy management problem for wireless mesh networks, JPEM-WMN). In questo caso, due tipi di stazioni d’accesso possono essere installate nella zona considerata: routers, in grado di connettersi ad altri dispositivi di rete, e gateways, che offrono anche un accesso diretto ad Internet. La domanda degli utenti situati nell’area varia a seconda di un profilo di traffico definito; inoltre, due gradi di congestione sono presi in esame, standard ed elevato. Il modello proposto è stato testato su sei scenari. Nonostante gli stessi vantaggi osservati per JPEM-CN si possano riscontrare nel caso delle reti a maglia, il risparmio energetico risulta minore a causa della scarsa flessibilità delle reti magliate in confronto a quella di reti cellulari eterogenee. Con piccoli incrementi delle spese di capitale, il modello JPEM-WMN permette comunque una riduzione del consumo di energia del 25–30% nella maggior parte dei casi. In questa tesi son esaminate alcune varianti dei due modelli di JPEM. Uno degli obiettivi del lavoro di ricerca è quello di valutare l’impatto di un sistema di modellizzazione congiunto rispetto ad un approccio più tradizionale in cui l’ottimizzazione delle spese di capitale è affrontata in un prima fase, seguita dalla gestione dell’operazione della rete installata. È stato quindi sviluppato un approccio in due fasi (two-step), in cui la rete di costo minimo è prima installata e successivamente gestita. I risparmi energetici misurati sono stati utilizzati per valutare la riduzione del consumo d’energia ottenuto tramite i modelli JPEM per reti cellulari e a maglia. Un’altra importante variazione, formulata per entrambe le tecnologie di accesso, consiste nel rilassare i vincoli di copertura globale al fine di garantire il servizio di rete soltanto per gli utenti attivi. In questo modo, le stazioni di base che coprono solo utenti inattivi possono essere spente, con l’effetto di diminuire ulteriormente la consumazione energetica della rete. I test condotti sugli scenari di rete cellulare dimostrano che, quando si considera una copertura di rete dei soli clienti attivi, è possibile ottenere un risparmio energetico supplementare del 25%–35%, mentre la percentuale è di circa 12% nel caso di reti magliate. Per quanto concerne unicamente le reti cellulari, questa tesi comprende una variante del problema JPEM-CN in cui il sistema non è libero di scegliere la topologia migliore (e quindi, gli investimenti di capitale) a seconda dell’importanza dei costi energetici nella pianificazione di rete. Al contrario, un parametro di budget è introdotto allo scopo di limitare i CapEx ad un dato valore. Il modello Budget JPEM-CN ottiene risultati simili alla versione originale; tuttavia, i nuovi vincoli sui costi di installazione della rete sembrano aumentare la complessità della formulazione. Riguardo le reti magliate, il modello On/Off JPEM-WMN presentato in questa tesi vincola i dispositivi d’accesso a cambiare il loro stato (da attivo a inattivo e vice versa) una sola volta nel corso della giornata. Questa formulazione ha come effetto la diminuzione dello spreco energetico durante la transizione di stato delle stazioni di base; tuttavia, il risparmio di energia derivato dallo spegnimento delle celle superflue alla copertura di rete è inevitabilmente ridotto. Altre variazioni minori sono state testate sul problema JPEM-WMN, come l’uso di una capacità variabile anziché fissa per i link della rete dorsale, o la simulazione del comportamento di una rete cellulare tramite l’installazione di soli gateways per eliminare la connettività multi-hop tipica delle reti a maglia. Al fine di ottenere più rapidamente una soluzione al problema congiunto per reti cellulari e permettere lo studio di scenari di test di più grandi dimensioni, nel corso della ricerca di dottorato è stato messo a punto un metodo euristico ad-hoc in cui i problemi di pianificazione e gestione di rete sono affrontati separatamente, così come i diversi intervalli temporali in cui è divisa la giornata. Partendo da una topologia completa, l’euristica trova il più efficiente pattern di attivazione delle stazioni di base in grado di soddisfare le esigenze di copertura sia durante il periodo di massimo traffico, quando è necessario servire un alto volume di traffico offerto, sia durante il periodo di minor traffico, quando un maggior numero di stazioni d’accesso può potenzialmente essere spento. Il pattern di attivazione risultante è considerato come una topologia parziale e fornito in ingresso al modello JPEM-CN originale. La topologia iniziale è quindi integrata per ottenere la miglior soluzione possibile valida per tutti i periodi temporali. Testato sugli stessi scenari di rete cellulare, il metodo euristico ha prodotto risultati che si discostano di circa il 10% dal limite inferiore dei risultati di JPEM-CN; inoltre, nuovi scenari di dimensioni realistiche sono stati risolti con successo in meno di 15 minuti nella maggior parte dei casi. Nel complesso, le diverse formulazioni di JPEM dimostrano che, quando la topologia di rete è concepita per essere eco-sostenibile, è possibile raggiungere risparmi energetici molto elevati al costo di un moderato aumento rispetto al minimo investimento di capitale. A questo scopo, è necessario tenere in considerazione gli effetti della gestione di rete durante la fase di pianificazione. I risultati illustrati e gli esempi numerici mostrano che la coesistenza nella stessa topologia di dispositivi d’accesso di diverse dimensioni è fondamentale per assicurare la flessibilità della rete e permettere l’adattamento della rete stessa alle variazioni di traffico nello spazio e nel tempo.
Radio planning and management of energy-efficient wireless access networks
BOIARDI, SILVIA
Abstract
In the last years, the information and communication technology (ICT) sector has transformed the way we live. Consistently delivering innovative products and services, the ICT assumed a primary role on economic development and productivity, becoming an integral part of everyday life. However, due to their wide and constantly increasing diffusion, the effect of information and communication technologies on global warming can no longer be ignored. The concept of Green ICT has originated with the aim of building awareness of this, thus boosting the research toward environmentally sustainable, energy-efficient technologies and solutions. As an important part of the ICT, telecommunication networks are experiencing a booming growth. Capacity issues and quality of service constraints are some of the main concerns that contribute to raise the power consumption. In particular, a large portion of the electricity bill results from the high power requirements of wireless base stations, which have been proved to be the most energy-hungry network components. Up to now, the mobile communication industry has focused mostly on the development of power-efficient mobile terminals, so as to attract a higher number of customers and con- sequently increase the operators’ profits; on the other hand, the research world has been investigating energy efficiency from a wider point of view. Besides studies on power-efficient devices and protocols, more recent works addressed the problem of energy-aware design and operation in wired and wireless network infrastructures. Many aspects and challenges of green network planning and management have been explored; nevertheless, the two problems have never been linked and tackled at the same time, neglecting the fact that an effective power-efficient network operation largely depends on the decisions taken in the design phase. The research presented in this doctoral thesis aims at filling this gap by developing an optimization framework that jointly considers the design and operation of wireless networks. The proposed joint planning and energy management problem (JPEM) strives to prove that, when cell sleeping is adopted as network management technique, the level of flexibility offered by the installed topology strongly improves the system capability to adapt to the varying traffic load. By minimizing the trade-off between capital expenditures (CapEx) related to the network deployment and operational and management expenditures (OpEx) calculated over the network lifetime, the model finds the most energy-efficient network topology while meeting the capital investment limitations imposed by the mobile operator. Two types of access network are analyzed in this work. As the name suggests, the joint planning and energy management problem for cellular networks (JPEM-CN) was developed to plan and manage the operation of cellular networks. Three cell sizes (macro, micro and pico) are allowed to be deployed in an area where no previous access devices are installed. A realistic daily traffic profile is assumed to characterize the area; therefore, the traffic offered by the test points randomly distributed in the area varies in each time period. Numerical results obtained by testing six different scenarios demonstrate that, with modest CapEx increases, a network planned to be energy efficient can reach power savings around 50%–60% compared to the energy saved by managing the operation of the minimum cost deployment. Moreover, the results give an interesting insight on the different topology compositions. While the minimum CapEx network is mostly composed of big and medium size cells, energy-aware topologies include a few larger cells in support of many small cells, which can be put to sleep during low traffic periods without leaving parts of the area uncovered. Besides JPEM-CN, another optimization framework has been modeled to solve the joint design and operation problem on wireless mesh networks (JPEM-WMN). In this case, only two types of access station can be placed in the area: routers, that can only connect to other routers, and gateways, having direct access to the Internet. Once again, a realistic traffic profile is considered, according to which the mesh clients vary their traffic requests; also, two degrees of traffic congestion are examined, standard and busy. The joint framework were tested on six scenarios. Even though the same benefits of the JPEM-CN apply to mesh networks, smaller energy savings are registered due to the lower flexibility of WMNs in comparison to heterogeneous cellular networks. However, the JPEM-WMN model yields savings of 25% to 30% in most cases with low extra installation costs. Additional model variants for both JPEM models are also presented in this research. Since one of the objective of the work is to evaluate the impact of using a joint modeling framework compared to a more traditional CapEx optimization and successive network management, a two-step approach is developed where the minimum cost network is first installed and then operated. The energy savings served to fairly evaluate the reduction of power consumption achieved by the cellular and mesh versions of JPEM. Another important variation, formulated for both types of access technology, consists in relaxing the total area coverage constraints to guarantee network service only to the active test points or mesh clients. This way, access devices having only idle users in their coverage radius can be put to sleep, further decreasing the energy consumption: results showed that supplemental power savings of 25%–35% can be reached in cellular test scenarios, while the percentage is around 12% in case of mesh instances. Referring now only to cellular networks, this thesis includes a JPEM-CN variant where the framework is not free to select the best topology (and thus, capital investments) according to the relative importance of the energy saving component in the network planning. Instead, a budget parameter is imposed in order to limit the CapEx to a certain value. The Budget JPEM-CN achieves similar results to the original version; however, the new constraints capping the capital costs seem to add complexity to the model formulation. For what concerns mesh networks, an interesting On/Off JPEM-WMN framework is presented where the access devices can switch their state, from idle to active and vice versa, only once per day, reducing the power wasted during state transitions but inevitably decreasing the energy savings from sleeping cells. Further minor changes have been also tested on the JPEM-WMN during the course of the doctoral research, as the introduction of variable backbone link capacity, compared to the fixed capacity in the original model, and the elimination of the multi-hop connectivity characteristics in order to simulate the behavior of a cellular network, by allowing only the installation of mesh gateways. In order to simplify the solution of the JPEM-CN and enable the evaluation of test scenarios closer to real size, an ad-hoc heuristic method was developed where the planning and operation problems, as well as the daily time intervals, are tackled separately. Starting from a complete topology, the heuristic finds the most efficient activation pattern that satisfies the coverage requirements for both the peak traffic periods, where high traffic volumes have to be served, and the off-peak, where the opportunity to save energy by turning off some access stations is the highest. The resulting activation pattern is considered as a new partial topology, which is provided as input to the original JPEM-CN; the initial topology is then enriched to obtain the best feasible solution for all time periods. For the same cellular test scenarios, the heuristic showed results only about 2% to 13% far from the respective JPEM-CN objective functions; on the other hand, new larger scenarios have been tested and successfully solved in less than 15 minutes in most of the cases. On the whole, the various JPEM formulations proved that higher energy savings can be obtained when the network topology is designed to be power efficient at the cost of moderate increases in capital costs. To do so, the effects of the network management have to be taken into consideration during the network design stages. Moreover, numerical results and examples showed how the coexistence of multiple sizes of access device in the same topology is fundamental to provide the network with enough flexibility to adapt to the traffic variations in time and space.File | Dimensione | Formato | |
---|---|---|---|
2014_07_PhD_Boiardi.pdf
accessibile in internet per tutti
Descrizione: Full thesis
Dimensione
2.56 MB
Formato
Adobe PDF
|
2.56 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/99707