Dense Mobile Communication Networks operating at 70GHz

Transcription

1 Dense Mobile Communication Networks operating at 70GHz October 2014 Yilin Li, Volker Pauli NoMoR Research GmbH, Munich, Germany Summary In recent years, mobile communication traffic has been increasing significantly and is expected to continue doing so in the future. On the one hand, current multiple access technologies and multi-antenna techniques improve the cell capacity, but are not able to meet the further increase of mobile data traffic. On the other hand, radio frequency bands of with wavelengths in the order of millimetres provide plenty of spectrum to satisfy the requirement of multigigabit data rates in future mobile communications. The unique propagation conditions of millimetre waves necessitate extreme beamforming and dense network deployment. However, connecting each access point with a wired link to the core network appears to be economically not viable in this deployment. Hence, in this paper we focus on a system using over-the-air in-band backhauling in which only a subset of the access points (AP) are equipped with wired backhaul while others connect to the core network via multi-hop relaying links. For such systems the deployment of an efficient algorithm for routing the traffic relayed between access points and exchanged between access points and user equipment is essential. In this paper, we propose such an algorithm. Systemlevel simulations illustrate the performance of the system as a whole and the benefits of the routing algorithm compared to a static routing approach. Introduction Mobile data traffic is projected to increase by 1000-fold by the year 2020 and probably ascend to times by the year 2025 [1]. In order to meet the growing traffic demand so far, 3GPP LTE Release 8 and 9 networks provide significant capacity improvements through flexible multiple access technologies and multi-antenna and interference mitigation techniques [2]. Capacity of LTE networks is further boosted through deployment of smaller cells with associated interference mitigation techniques based on the 3GPP LTE Release 10 and 11 [3]. In spite of all the improvements, the times increase of mobile communication traffic cannot be accommodated unless additional bands at very high frequencies are allocated to mobile communication networks. Due to the encouraging propagation features for radio communications, a large number of commercial radio communication applications including AM/FM radio, GPS, and Wi-Fi have been crowded in a narrow band of the Radio Frequency (RF) bands from 300 MHz to 3 GHz, which are widely known as the sweet spot. On the contrary, the RF bands from 3 GHz to 300 GHz, where the corresponding wavelengths are in the scope of millimetre wave (mmw), have been widely unused for mobile communication applications. One of the key advantages of mmw mobile communication technology is hence the abundance of available spectrum. For example, RF bands of GHz, GHz and GHz ( E-band ) could offer in total 12.9 GHz Nomor Research GmbH / info@nomor.de / / T /9

2 bandwidth, which is far more than that of all other licensed lower RF bands so far. However, adverse propagation conditions bring challenges to mmw communication systems. In particular, electromagnetic waves at this frequency propagate in a way very similar to visible light; so even a user's body will when located between the AP and his mobile device severely attenuate the signals exchanged between UE and AP. Furthermore, propagation attenuation of mmw due to atmosphere absorption of oxygen molecules and water vapour is significantly higher than that of lower frequency bands [4]. When combined with the small aperture of typical antennas for mmw applications these special propagation conditions consequently call for large antenna arrays, for extreme beamforming and for very dense deployments. Recent research [6]-[9] demonstrates that mmw is practical for mobile communications, especially for small size networks. In this work, we focus on such very dense (intersite distance in the order of 100 m) networks with access points (AP) operating at 70 GHz with a system bandwidth in the range of 1 GHz to 2 GHz. Very dense deployments bring the constraint that equipping each AP with its own wired backhaul appears to be economically not feasible. Therefore, it is deemed to be attractive to only equip a subset of the APs with the wired backhaul to the core network while others are connected to the core network via (multi-hop) relaying from the wired backhaul-aps. For maximum flexibility in resource allocation, the relaying shall be performed in the same spectrum as the communication between APs and UEs. This imposes additional challenges in terms of routing the user traffic between a wired AP and the AP that is directly communicating with a given UE. In this work, we propose an efficient routing algorithm that achieves high system performance of the over-the-air backhaul networks. While the relay radio access technology itself is not the subject of this paper, in-band relays have already been standardized in LTE-Advanced Release 10 [10]. The remainder of this paper is organized as follows: Following the above brief introduction of Millimetre wave and its benefit as well as challenges in mobile communications a communication system with a Manhattan-like grid network of access points and multi-hop relaying links is described in the following section. Subsequently, this paper proposes an efficient routing algorithm for multi-hop relaying links. At this, we simulate two route selection schemes at different densities of backhaul-aps and network loads and discuss the conclusions from the simulation results. System Overview Network Layout We consider a mobile communication system deployed in an urban street canyon environment (Manhattan Model) as it is commonly considered for mmw applications [6]. In the network, the streets are 150 meters long and 20 meters wide. Configurations of APs and an exemplary distribution of UEs are illustrated in Figure 1. At every crossroad there is an AP; the locations of these 36 APs are marked by the black circles in Figure 1. Some of the APs will be equipped with wired backhaul, where in Figure 1 the number of backhaul-aps is 4; these are marked using additional black crosses inside the outer circles. UEs are randomly dropped in the street canyons; their locations are given by the small blue crosses. Due to the light-like propagation characteristics of mmw signals, a signal traveling between an UE and an AP is severely attenuated, if the line of sight (LOS) between both entities is blocked by a building (black square in Figure 1) [1]. As a consequence, those UEs that have no direct LOS connection link to one of the backhaul-aps will typically rather pick another AP, namely the one that provides the strongest access link, to connect to and through which to communicate with one of the backhaul-aps via a multi-hop relaying link. In Figure 1, the black arrow Nomor Research GmbH / info@nomor.de / / T /9

3 represents an example of a direct connection link between an UE and an AP with wired backhaul, while the red arrows represent an example of a three-hop relaying link between both entities. role in such a system. Resource allocation: APs operate in halfduplex mode. So, for backhaul-aps all radio resources can be allocated to serve other APs as backhaul or to serve UEs directly. For other APs, the radio resources will have to be shared between relaying towards the AP with wired backhaul and serving as backhaul to other APs or serving UEs directly. As we mentioned in the above section, routing the user traffic between a wired AP and the AP that is directly communicating with the given UE is regarded as the key factor to achieve high system performance. For such a system, we designed a routing algorithm, which is described in the following section. Figure 1: Network Layout Assumptions Our systems are built based on the following assumptions: Line-of-sight (LOS) conditions: We assume that the communication between APs and also between AP and UE are always in the LOS conditions unless LOS is obstructed by a building. Each UE will connect to the AP that provides the strongest link in terms of minimal pathloss and from there potentially i.e. if that AP does not have a wired backhaul itself to one of the backhaul-aps via multi-hop relaying links to ensure the LOS conditions. In-band relaying: The carrier frequency of communications between APs is the same as that used for communication between AP and UE. Propagation attenuation: Our system uses extreme beamforming to achieve a sufficiently high rise over thermal. As a consequence, interference plays a minor Routing Algorithm Design Our routing algorithm aims at finding the best route for each UE when it becomes active and hence requires a multi-hop relaying link. It does so in a greedy manner, i.e. taking into account but not changing the routes already allocated to other UEs that are already active, i.e. that are already transmitting or receiving data. Optimality is defined in terms of maximum data rate achievable considering the SINRs and availability of resources on each hop of a potential route. When a UE s session has ended its route is torn down and a new one, possibly different from the previous one, is established whenever the UE becomes active again. In the algorithm, we abstract the network as a graph, where the nodes and edges in the graph represent the network elements (APs and UEs) and the communication links between two adjacent network elements, respectively. Each edge is associated with a weight, which is taken as the achievable data rate between the two corresponding adjacent nodes. In the following the three main procedures of our routing algorithm are described. Graph Search Algorithm The route of highest rate is found by a graph Nomor Research GmbH / info@nomor.de / / T /9

4 search algorithm in the graph defined above. This algorithm is able to find a path between two nodes in a graph where the minimum weight of all the edges within the path is larger than that of other possible paths between these two nodes. The motivation of this graph search algorithm is based on the fact that the actual achievable rate for a multi-hop link is constrained by the weakest hop of which the data rate is the lowest. If we want to improve the data rate of a path, we have to increase the lowest one of all the hops. Thus, as long as the edge weights are set as the achievable data rates of the corresponding hops and the candidate metric of a node is defined as the minimum of the rate between this node and a given predecessor node and the metric of that predecessor node, a path with highest data rate can be obtained by implementing fairly standard graph searching algorithms between an UE node and a backhaul- AP node. Route Selection Scheme One should notice that there could be more than one backhaul-ap in the network, therefore each UE could communicate with any of them. Therefore, multiple paths would be found between different backhaul-aps and a considered UE, and each of these paths can provide a particular highest end-to-end data rate. Obviously, we should select the one with the maximum end-to-end data rate among these paths / backhaul APs. Update Network After the instantaneously optimum path is determined for an UE, the actual data rate of this path is obtained as being equal to the lowest data rate of all hops of that path. As the data rate of the path cannot be further improved if the weakest hop is settled, there is no need for other hops to offer higher data rates than the lowest one. Therefore, after the path has been found for each UE, our routing algorithm will calculate the amount of radio resources needed to achieve the lowest data rate, and the information about the network is updated in terms of the new achievable data rates and remaining allocable radio resources of all the hops in the network. This concludes the update of the routing and the thus found routes will not be changed until a user's data session ends. At this time, the radio resources allocated to him will be released and a new route will be designated to this user only when he re-accesses the network for his next data session. System-Level Simulation Simulation Parameters We simulated a scenario as illustrated in Figure 1 with 36 APs; 1, 2, 4 or 8 of them are equipped with a wired backhaul into the core network. The whole simulation area is divided into several equally small pieces according to the number of backhaul-aps. Each of them is located at the centre of such a small area. 100 UEs are uniformly and randomly dropped in the streets. Other key parameters of the simulation are listed in Table 1. We point out that being implemented in a dynamic TDD system the scheme can be agnostic as to whether the traffic is transmitted in the uplink or downlink direction. Parameter Bandwidth Carrier frequency Traffic type AP TX power Noise power Max. beamforming gains Value 2 GHz 72 GHz FTP 30.8 dbm dbm AP:18 dbi UE: 6 dbi Table 1: Simulation Parameters Simulation Scenarios In the previous section we described the dynamic route selection scheme. As a reference, another route selection scheme, which is herein referred to as static routing, is also implemented. In this selection scheme, fixed routes are assigned to each AP and hence UE to Nomor Research GmbH / info@nomor.de / / T /9

5 connect to the nearest backhaul-ap and the graph search algorithm will not be carried out in this scheme. example can only provide a data rate of 3 as illustrated in Figure 3. We simulated both these route selection schemes and compared their performance in the following subsections. Simulation Results In the following, we present some simulation results to illustrate the effect of our routing algorithm. We ran simulations with different numbers of backhaul-aps and for different network loads. Figure 2: Dynamic Routing Figure 2 illustrates the dynamic routing scheme. In this figure, four backhaul-aps are represented by gray circles while the AP connecting to the UE is represented by red circle. Moreover, numbers in gray circles indicate the achievable data rates of the links from the respective backhaul-aps. As dynamic routing will select the route that achieves the highest data rate, in this figure the UE will connect to the backhaul-ap that offers the data rate of 8. Results are illustrated by plotting the average session throughputs and edge session throughputs of all users vs. the offered load in Figures 4 and 5, respectively. Here, the edge session throughputs are defined as the fifth percentile of the session throughputs distribution. The offered load is defined as the total amount of data that the network delivers to all users divided by the simulation time, in other words: the average data rate transmitted throughout the network. Markers indicate actual measurements between which we interpolate linearly with lines. The idea of this key performance indicator (KPI) is: At a given offered load, the user experience is better if the corresponding curve lies higher; or the same user experience can be delivered to more users if the curve lies farther to the right. Figure 3: Static Routing On the contrary, when using the static routing with predefined fixed routes the UE will connect to the nearest backhaul-ap, which in this From the figures we can observe that, on the one hand, for a certain amount of offered load the curves move higher with the increase of the number of backhaul-aps, i.e. the more backhaul- APs there are, the better the user experience will be. More backhaul-aps lead to more allocable resources therefore users experience higher data rates. While this increase is almost linear in the number of backhaul-aps when going from 1 to 2, it is less for even larger numbers of backhaul- APs as the network is more and more loaded with relayed traffic. Nomor Research GmbH / info@nomor.de / / T /9

6 approximately 17Gpbs to 23Gbps for a single backhaul-ap. Furthermore, the dynamic routing reduces the gap between average and edge session throughput, hence improving the fairness of the system. The average route lengths over all users are illustrated in Figure 6. X-axis indicates the maximum number of backhaul-aps and y-axis indicates the average route length. At each backhaul-aps case (max. numbers equal to 1, 2, 4, or 8), the bars are plotted in the order of decreasing network load along the positive direction of x-axis. Figure 4: Average session user throughput Figure 6: Average route length Figure 5: 5%-tile session user throughput Secondly, the dynamic routing exhibits drastic performance improvements compared to the static routing. Here, both average and edge session throughputs of dynamic routing are significantly higher than those of static routing. In particular, edge session throughputs of dynamic routing with maximum 2 backhaul-aps are higher than the static routing with 4 backhaul-aps and are even close to that with 8 backhaul-aps. The offered load, at which the percentage of utterly unsatisfied users reaches 5% is increased considerably, e.g. from From this figure we can find longer average route length in dynamic routing than static routing. Due to the flexibility of backhaul-ap selection, the network could choose to connect a UE to a backhaul-ap that is far from it but offers a higher data rate. This way, network load can be more evenly distributed over different backhaul-aps and also different intermediate hops. Moreover, along with the increase of the number of backhaul-aps, the average route lengths of both schemes decrease. With more backhaul- APs, UEs dropped in the border of the network could (dynamic routing) / have to (static routing) Nomor Research GmbH / info@nomor.de / / T /9

7 communicate with the nearest backhaul-ap compared to the case that UEs must go through long routes to reach the single backhaul-ap. Conclusions In this paper we have briefly summarized the benefits of and some of the challenges in millimetre wave mobile communications. We proposed an efficient routing algorithm for overthe-air backhaul for dense mobile communication networks. We provided some system-level simulation results illustrating the great potential of mmw communication systems as a whole to achieve very high average data rates (more than 1 Gbps). Furthermore, we showed that a dynamic routing scheme can offer significantly better system performance than a simple static routing scheme. References [1] Larew, S.G.; Thomas, T.A.; Cudak, M.; Ghosh, A., "Air interface design and ray tracing study for 5G millimeter wave communications," Globecom Workshops (GC Wkshps), 2013 IEEE, vol., no., pp.117, 122, 9-13 Dec [2] 3GPP TS , "Evolved Universal Terrestrial Radio Access (E-UTRA) and Evolved Universal Terrestrial Radio Access Network (E- UTRAN); Overall Description; Stage 2," June 2009 (R9-2009). [3] 3GPP TS , "Evolved Universal Terrestrial Radio Access (E-UTRA) and Evolved Universal Terrestrial Radio Access Network (E- UTRAN); Overall Description; Stage 2," December 2011 (R ). [4] Zhouyue Pi; Khan, F., "An introduction to millimeter-wave mobile broadband systems," Communications Magazine, IEEE, vol.49, no.6, pp.101, 107, June 2011 [5] Marcus, M.; Pattan, B., "Millimeter wave propagation; spectrum management implications," Microwave Magazine, IEEE, vol.6, no.2, pp.54, 62, June 2005 [6] Hong Zhang; Venkateswaran, S.; Madhow, U., "Channel Modeling and MIMO Capacity for Outdoor Millimeter Wave Links," Wireless Communications and Networking Conference (WCNC), 2010 IEEE, vol., no., pp.1, 6, April 2010 [7] Ben-Dor, E.; Rappaport, T.S.; Yijun Qiao; Lauffenburger, S.J., "Millimeter-Wave 60 GHz Outdoor and Vehicle AOA Propagation Measurements Using a Broadband Channel Sounder," Global Telecommunications Conference (GLOBECOM 2011), 2011 IEEE, vol., no., pp.1, 6, 5-9 Dec [8] Kyro, M.; Ranvier, S.; Kolmonen, V.; Haneda, K.; Vainikainen, P., "Long range wideband channel measurements at GHz frequency range," Antennas and Propagation (EuCAP), 2010 Proceedings of the Fourth European Conference on, vol., no., pp.1, 5, April 2010 [9] Azar, Y.; Wong, G.N.; Wang, K.; Mayzus, R.; Schulz, J.K.; Hang Zhao; Gutierrez, F.; Hwang, D.; Rappaport, T.S., "28 GHz propagation measurements for outdoor cellular communications using steerable beam antennas in New York city," Communications (ICC), 2013 IEEE International Conference on, vol., no., pp.5143, 5147, 9-13 June 2013 [10] 3GPP TS , "Evolved Universal Terrestrial Radio Access (E-UTRA); Physical Layer for Relaying Operation," October 2010 (R ). Note: This white paper is provided to you by Nomor Research GmbH. Similar documents can be obtained from Feel free to forward this issue in electronic format. Please contact us in case you are interested in collaboration on related subjects. Please note in our assessment(s) we only considered those facts known to us and therefore the results of our assessment/assessments are subject to facts not known to us. Furthermore, please note, with respect to our assessment(s) different opinions might be expressed in the relevant literature and for this purpose there may be some other interpretations which are scientifically valid. Nomor Research GmbH / info@nomor.de / / T /9

8 System Level Simulation Services Nomor Research has developed a comprehensive simulation environment supporting various standards such as LTE, LTE Advanced and HSPA and offers related services to support research, development and standardisation. Features of the dynamic multi-cell, multi-user system level simulator include: macro-cell and HetNet deployments (pico-, femto-cell, relay nodes) flexible base station and user configurations and drop models different transmitter and receiver chains incl. MIMO, ZF, MMSE channel modelling with slow/fast fading, pathloss, full user mobility intra- and intercell interference modelling for OFDMA, SC-FDMA and WCDMA 2D and 3D antenna pattern and multiantenna beam forming extensive metrics and KPIs: capacity, throughput, spectral efficiency, user QoS etc The simulator can be used on project basis or in customized simulation campaigns. The performance of the system level simulator has been calibrated to simulation results obtained in standardisation. Research on advanced algorithms include, but are not limited to: various aspects of scheduling and resource allocation algorithms considering channel and buffer status, QoS etc. inter-cell interference coordination, avoidance and cancellation single user-, multi-user MIMO with open and closed loop feedback cooperative multi-point transmission and reception functions for self-organising and selfoptimizing networks (e.g. load balancing, mobility optimization, tilt optimisation, range extension, power saving etc.) If you are interested in our services please contact us at info@nomor.de or visit us at Nomor Research GmbH / info@nomor.de / / T /9

9 Standardisation Service Standardisation is often essential to drive technology, to get knowledge about market trends, customer or competitors and basically to lay ground for future business with the required knowledge for implementation. Real impact in standardisation will be time and cost intensive since it requires years of attendance with excellent contributions, working across various groups and contacts to the decision makers, and is thus mostly limited to the key player. Nomor Research's standardisation services provide you the resources and the knowledge needed to understand and influence standardisation. Sharing the resources between partners and projects maximized your impact and presence while limiting your cost. 3GPP RANx, 3GPP SAx, ETSI, DVB, IETF, ITU, ISO/MPEG, and DASH-IF, we can support you in manifold tasks on short notice as well as in long lasting projects. Contact us at or visit Consulting can include, but is not limited to: Regular standardisation updates, In-depth information on technical areas, Release overview, analysis, feature roadmaps and complexity analysis, Contribute and influence standards activities, Represent your company at standardization, Analyze the feasibility of concepts for standardization, Evaluation of own and other contributions, Answer questions concerning certain standards. Technology Training NoMoR provides you professional training held by distinguished expert in mobile communication industry. Standard Courses: LTE Technology Training LTE Signalling/Protocols LTE MIMO Technologies LTE-Advanced LTE Self-Organizing Networks LTE Heterogeneous Networks As development cycles are getting shorter and new technologies are emerging frequently in a rapidly changing market, your teams need to act quickly once strategic decisions have been taken. Professional training tailored to your specific needs and any level of background will get your team up-to-date and will save you money, since your staff can focus on the task to be done. Contact us at training@nomor.de or visit Nomor Research GmbH / info@nomor.de / / T /9