Heterogeneous All-IP Wireless Broadband with WiMAX, WiFi and HSPA

Heterogeneous All-IP Wireless Broadband with WiMAX, WiFi and HSPA Pål Grønsund, Arild Jacobsen, Tor Ove Breivik, Vegard Hassel, Geir Millstein, Thomas...
Author: Claribel Holmes
0 downloads 1 Views 598KB Size
Heterogeneous All-IP Wireless Broadband with WiMAX, WiFi and HSPA Pål Grønsund, Arild Jacobsen, Tor Ove Breivik, Vegard Hassel, Geir Millstein, Thomas Haslestad Telenor R&I, Snarøyveien 30, 1331, Oslo, Norway {pal.gronsund, arild.jacobsen, tor-ove.breivik, vegard.hassel, geir.millstein, thomas.haslestad}@telenor.com Abstract—Wireless network operators often offer wireless broadband services by using different wireless technologies such as WiMAX, WiFi and HSPA. Each wireless technology often uses a separate and stand alone network architecture, and thus operations and management may easily become a complex and expensive task. In this paper we present the design and implementation of an all-IP heterogeneous network where the services provided are public and private wireless broadband access based on Fixed WiMAX, WiFi and HSPA. These basic services and combinations of these are further used to design different commercial products in the network. Radio coverage measurements are then performed for these technologies in interesting parts of the realized deployed network. The overall network implementation has been in successful operation for several months. Keywords-Wireless; WiMAX; WiFi; Hetereogeneous; Backhaul; Architecture;

I.

HSPA;

Broadband;

II.

THE HETEROGENEOUS ALL-IP NETWORK ARCHITECTURE

The access technologies WiMAX (IEEE 802.16d-2004 [1]), WiFi (IEEE 802.11b/g [2]) and HSPA are deployed in a heterogeneous all-IP network architecture including an aggregation point and core network. As illustrated in Figure 1, the following scenarios are deployed by using: •

HSPA as mobile broadband technology



HSPA as backhaul for WiFi hot-spots



WiMAX as backhaul for WiFi hot-spots



WiMAX as wireless DSL-replacement



WiFi mesh for connecting local AP’s

INTRODUCTION

Today, most laptop computers and many cellular phones and camera devices are WiFi enabled. The high number of WiFi enabled terminals constitutes a substantial opportunity for operators. At the same time, operators investing in wide area wireless broadband technologies like UMTS/HSPA and WiMAX often subsidize their terminals in order to rapidly increase the terminal penetration in the population. Hence, there should be an interest for such operators to use existing HSPA and WiMAX infrastructure also as a low-cost backhaul solution for WiFi hot-spots, in order to extend their services to more terminals and thereby increase the operators’ footprints. In this paper, we describe a commercial implementation combining WiFi, WiMAX and HSPA into one heterogeneous all-IP network. Since each technology has its own advantages and characteristics, this enables the operator to select the access technology constituting the most favorable solution in each particular situation. The integration of WiFi, Fixed WiMAX and HSPA into the same network using the same IS/IT core network has to our knowledge not been realized before. In this paper we first briefly present the network architecture that was implemented. Then, in Section III, we outline the services that are currently offered in the network. Further, in Section IV, we go into detail on how the access network and the backhaul were implemented. In Section V we describe our findings from different technical tests that were performed in the network. Finally, we summarize our key findings in Section VI.

Figure 1 Network Architecture

Traffic from the access networks is transported to the BRAS (broadband remote access server) responsible for rejecting or allowing access to the Internet. This is done by communicating with the core network and the AAA (authentication, authorization and accounting) server by using the RADIUS (remote authentication dial in user service) protocol. Services are controlled and provisioned at the BRAS. WiMAX and WiFi are connected directly to the BRAS over Ethernet and HSPA uses tunneling to establish an Ethernet

connection to the BRAS. Since the HSPA network is just used to carry Ethernet packets, the IS/IT core of the HSPA network is not integrated into the architecture of our heterogeneous allIP architecture. Operational and management services such as captive portal, customer self care and network monitoring residing in the core network are shared across all access networks. This means that a user will be indifferent to the type of access network. III.

SERVICES AND PRODUCTS OFFERED

The services provided over the deployed network are public and private wireless broadband access over WiMAX and WiFi. The public service is typically granted through a prepaid subscription and the private service through a postpaid subscription. HSPA mobile broadband services are private, both postpaid and prepaid. Various wireless broadband products may be designed as combinations of the services and wireless technologies illustrated in Figure 2.





Public and private WiMAX-WiFi (public and private services combined) o

Public service by WiFi backhauled by WiMAX

o

Private service by WiMAX and WiFi

Private WiMAX (private service by WiMAX)

Summarized, the public access services are generally granted through a prepaid WiFi subscription and private access services are generally granted through a postpaid WiMAX, but also WiFi, subscription. IEEE 802.1Q VLAN (virtual LAN) tagging is used to differentiate the services in BRAS, which is illustrated by the lines in Figure 2. IV.

IMPLEMENTING ACCESS AND BACKHAUL NETWORKS

A. Fixed WiMAX Backhaul for WiFi The network infrastructure is built to support VLAN tagging from the outer points in the access network, through the aggregation network and into the core network. The traffic flow in the WiMAX-WiFi network is illustrated in Figure 3 for the solution with public and private services combined at one WiMAX-WiFi site. VLAN tagging is used to separate the different public and private services, where one VLAN tag is used for the public service and another tag is used for the private service. Management traffic also uses a separate VLAN tag “Mgmt”.

Figure 3 Traffic flow in the WiMAX-WiFi network. Public service (blue packets), and private service (red packets) and management traffic (white packets) are differentiated by using separate VLANs

Figure 2 Services (public in blue, private in red, HSPA in black) and products (HSPA, public HSPA-WiFi, public WiMAX-WiFi, public and private WiMAX-WiFi, private WiMAX)

Starting at the top in Figure 2, the wireless broadband products (i.e. combination of one or more services and wireless technologies) can be named and described as: •

HSPA



Public HSPA-WiFi (public backhauled by HSPA)



Public WiMAX-WiFi (public service by WiFi backhauled by WiMAX)

service

by

WiFi

It should be noted that VLAN tagging is not done by the public users or private subscribers. VLAN tagging for the public service is done by the WiFi access point (AP) and VLAN tagging for the private service is done by the WiMAX consumer premises equipment (CPE). Quality-of-service (QoS) is configured at two points for the WiMAX CPEs. Firstly, two WiMAX QoS profiles are configured in the WiMAX network management system (NMS) system for appliance in the WiMAX base stations (BSs), one for the public connection and a one for the private WiMAX connection. Secondly, QoS profiles with maximum rate restrictions for the public WiFi users are configured in the operations subsystem (OSS) in the core for appliance in the BRAS.

B. Fixed WiMAX (“Wireless DSL Services”) Traffic for private services by Fixed WiMAX uses PPPoE (point-to-point protocol over Ethernet) connections, and different VLAN allocations may be provided based on endcustomers, customer groups, city or regions. The PPPoE connections are terminated in the BRAS. C. WiFi Mesh for Locally Connecting WiFi APs Both indoor WiFi mesh solutions in hotels and outdoor WiFi mesh solutions at hotels, marinas and public areas have been deployed in the network. The WiFi mesh technology enables the APs to communicate wirelessly between each other so that not all APs in the network need backhauling from the WiMAX network. The APs that have WiMAX backhaul are denoted gateways. The indoor mesh solution communicates at 2.4 GHz both between the mesh APs and between the APs and the customers, whereas the outdoor mesh solution uses dual-radio mesh with 2.4 GHz for customer access and 5 GHz between the mesh APs. The outdoor solution therefore requires line-of-sight (LOS) between the mesh nodes. Obtaining LOS is also easier in an outdoor environment compared to an indoor environment. D. HSPA as Mobile Broadband Technology Ordinary mobile broadband HSPA users uses the regular HSPA core network for service control and other required functionality, and the HSPA network is in this respect therefore not part of the all-IP network architecture. Services control and provisioning is carried out in traditional mobile networks. E. HSPA as Backhaul for WiFi Hotspots A solution for public WiFi service delivery for HSPA operators was developed. This solution was realized by means of tunneling traffic from the HSPA-WiFi access router through the ordinary mobile network to the router in the aggregation network as illustrated by the tunnel in Figure 2. The result is that WiFi sessions from the HSPA access network appear similar to the aggregation network as WiFi sessions from the WiMAX-WiFi access network. WiFi users connecting to the HSPA-WiFi routers can therefore be identified and charged as individual WiFi users by the core network, and not as in the normal, un-tunneled case where the HSPA-WiFi router is identified and charged by the mobile core network as one single HSPA user with a SIM card. Layer 2 tunneling protocol (L2TP) [3] was used for the implementation of this solution, where L2TP supports Ethernet traffic as in the WiMAX and WiFi networks. Ethernet traffic is then supported from the end terminals throughout the network towards the core network and the Internet. The solution enables the HSPA-WiFi device-to-aggregation network connection to re-use the existing mobile core and transport network for public WiFi service delivery. This provides all WiFi users with the same interface independent of backhaul technology. The solution can be applied for any WiFi service platform, which typically could be in situations where operators have HSPA coverage but where DSL backhaul is not available.

QoS settings have also to be set for the HSPA connection in the HSPA access network. Then, QoS profiles for the HSPA connections in the heterogeneous all-IP network are configured in the OSS in the core, which actually is bitrate setting for the L2TP tunnel. The WiFi QoS profiles are already configured as explained in the WiMAX-WiFi service for the private WiFi sessions. V.

MAIN FINDINGS FROM TECHNICAL TESTS

A. WiMAX Coverage Basic quality parameters for the Fixed WiMAX radio links have been forecasted and measured in the network described. A simple analysis has been carried out on these data, concluding that they are reasonably well aligned with the forecast. A high domination of the highest modulation (64 QAM) is found to be used in the network, with just a few exceptions. This indicates that most links in a Fixed WiMAX network with distances between the BS and the CPEs up to 5 km will operate on the maximum available modulation. The measurements show that the uplink is highly stable with respect to the carrier-to-interference-plus-noise ratio (CINR) and received signal strength. This stable performance may be caused by automatic transmission power control (ATPC) in the CPE. The same measurements for the downlink show instability and the reason for this behavior should be investigated further. The relationship between CINR on one side and preferred modulation type and error correction code rate on the other side should also be further investigated. Finding this relationship will be very useful when new Fixed WiMAX networks are to be planned. Generally, Fixed WiMAX provides higher average throughput in a cell compared to HSPA mainly due to better and more stable link conditions in general Fixed WiMAX deployments. Hence, Fixed WiMAX is considered to be better suited to support high capacity customers in fixed locations. This is the normal case where HSPA dongles are used. We have also performed measurements with external antennas for HSPA, as described below. B. Indoor and Outdoor WiFi Coverage Two different WiFi network architectures were investigated at two different hotels: •

Outdoor WiFi APs providing indoor coverage: several buildings spread with 250 m from one end to the other, 500 rooms, 4 APs



Indoor WiFi APs providing indoor coverage: a 80m x 60m building complex with open inner rooms, 124 rooms, 7 APs

Indoor WiFi deployments gave in general very limited signal strength close to the façade/corners of the building, but good coverage for rooms with doors facing the open inner areas where the APs were placed. About 50% of the rooms got service coverage, i.e. about 10 rooms per AP.

Indoor coverage obtained from outdoor deployed APs gave a good solution when the APs were mounted up to about 80 m from the building, see Figure 4. Based on measurements in a number of rooms at second floor and randomly picked, we estimate that 70% of the rooms facing the open area between the buildings will have acceptable service coverage (see Figure 5). When the RF signals’ angle of incidence was more than about 45 degrees to the building façade, the signal degradation seen from inside of the building started to be significant.

Figure 5 Resulting WiFi signal coverage

C. HSPA Coverage with External Antennas HSPA [4] measurements were done in an urban environment with a high density of BSs. Consequently, it was likely that this was an interference limited area. One typical test site had 6 BSs within 400 m distance and 28 BSs within 800 m distance from the apartment used for measurement. An HSPAWiFi router was used with three different external HSPA antennas. Received signal strength, signal-to-interference ratio and throughput were measured inside two apartments. Theoretical chip energy, Ec to interference, I0 is given by Figure 4 Hotel area with 4 external WiFi APs

At the investigated sites, a combination of indoor and outdoor deployment of APs would be preferable. The tests indicate that when the building façade have a certain transparency for radio signals, in many cases it will be faster and more cost-efficient to deploy outdoor WiFi networks in order to obtain indoor coverage. In the test case, the service per user was rate limited to 1 Mbit/s for the downlink and 512 kbit/s for the uplink. The tests showed that when the CINR reached the level where a stable service was possible, the user got throughputs similar to the rate limitation set in the network. Consequently, the high throughput potential of IEEE 802.11g was not exploited.

Ec PD αPD , = = I0 P D + PI + PN αPD + αPI + PN where PD is the desired signal power, PI is the power received from interfering sources, PN is the background noise, and α=f(d,l) where d is the distance from the apartment’s window wall into the apartment and l is antenna cable length. Figure 6 shows the signal power measured for one apartment.

HSPA-WiFi router downlink throughput significantly. The reason for this is that a directional antenna will both increase the received power and decrease the interference from other BSs. This conclusion is only valid if the feeder cable between the HSPA-WiFi router and the antenna is not too long.

Received Signal Code Power -75

RSCP [dBm]

-80 -85 BS 41732

-90 -95 -100 -105 0

2

4

6

8

10

12

Distance [m]

Figure 6 Measured received signal from two different BSs as function of distance from the street window

When signal level is significant above the background noise, i.e. α(PD+PI)>>PN, the coverage quality indicator can be simplified to

Ec PD ≈ I0 P D + PI . To verify this assumption for the given measurement site, the relation was measured, Figure 7. Chip energy to Inteference 0 -2

Ec/Io [dB]

VI.

BS 41708

-4 -6

BS 41732

We have implemented an all-IP heterogeneous network which we consider to be the future network infrastructure, where service functionality is independent of radio access technology. Hence, any access technology that supports IP can be used together with the supported services. The access technologies WiMAX and WiFi carrying Ethernet links have been successfully used for major parts of the network. HSPA does not use Ethernet and does therefore not naturally plug into the heterogeneous all-IP network architecture. However, a solution was implemented and deployed for public WiFi access backhauled by HSPA by means of tunneling. This tunneling solution may also be used by any access technology that not natively runs on Ethernet technology. The key benefit of the heterogeneous all-IP network architecture is the flexibility. Firstly, the network operator may use any access network that supports IP at any time. Secondly, the network uses standardized protocols and is therefore independent of equipment vendor in all parts of the network. Consequently, costs of deploying and operating the network can be reduced significantly and wireless broadband products of high quality can be offered at lower costs. The key technical learnings from the network implementation described in this paper can be summarized as follows:

BS 41708

-8



A grand majority of the Fixed WiMAX CPEs that were deployed within 5 km from the BS had the highest modulation (64 QAM)



Providing indoor WiFi coverage at a hotel where the rooms are facing common façade will be most efficiently be covered by outdoor WiFi access points

-10 -12 -14 0

2

4

6

8

10

12

Distance [m]

Figure 7 Variation of chip energy to interference related to intrusion depth into the apartment

The service functioned with satisfying throughput until the received signal dropped below -91 dBm. The main conclusion was that in an urban area with a high density of BSs, an external omni-directional HSPA antenna mounted on a window will in most cases not improve the throughput of the HSPAWiFi router. The reason for this is that in an interference limited network, both the signal and the interference will increase when the antenna connected directly to the unit is replaced by an external antenna placed closer to the window using cable. Consequently, it will not increase the signal-tointerference ratio, and hence neither the downlink throughput. On the other hand we found that in an urban area with a high density of BSs, an external, directional HSPA antenna mounted outdoors, pointing towards the BS, will improve the

CONCLUSION

For an interference limited HSPA network, using external omni-directional HSPA antennas for a HSPA-WiFi router will in most cases not improve the throughput. However, using a directional antenna pointing in the direction of a BS will improve the throughput of such an HSPA-WiFi router REFERENCES [1]

[2]

[3] [4]

IEEE Std 802.16d-2004 “IEEE Standard for Local and Metropolitan Area Networks Part 16: Air Interface for Fixed Broadband Wireless Access Systems”, June 2004 IEEE Std 802.11g-2003 “Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications”, LAN/MAN Standards Committee of the IEEE Computer Society, June 2003 IETF, “Layer Two Tunneling Protocol L2TP”, RFC 2661, August, 1999 3GPP, “TR 35.858 Physical Layer Aspects (Release 5)

Suggest Documents