This “Product How-To” article focuses how to use a certain product in an embedded system and is written by a company representative.
Modern mobile devices are capable of transferring an increasing varietyof data types that continue to grow in complexity. For example, mobilephones, once used only for transmitting/receiving voice data, are nowcapable of playing and sending music files, capturing and transmittingimages, and even recording and sending video clips.
Moreover, they can be used as wireless modems that connect PCs tothe Internet. As the designs of these devices get more and moreintricate, it becomes increasingly important that engineers test theperformance of each aspect involved in the data transfer.
|Figure1. There are six subsystems involved in an HSDPA/W-CDMA network.|
Designers of mobile devices typically test the efficiency of datatransfers from baseband data analysis. A better scenario is to test thedevice in a real network environment. However, there are three issuesinvolving mobile networks during test.
First, the network – for instance, a high-speed downlink packetaccess (HSDPA) network or W-CDMA – may not be available.Second, if thedevice is connected to a real network and a problem arises (e.g. a linkdisconnection or a slow down in throughput), it is very difficult totrace the actual cause in any of the subsystems.
Also, the real network itself varies over time, and the quality itprovides to user equipment (UE) is not guaranteed and reproducible.Thus, performance evaluation of the UE is not reliable while connectingit to a real network.
Figure 1 above shows the sixsubsystems involved in an HSDPA/W-CDMA network: the UE, Node B (basestation), radio network controller, 3G-Serving GPRS Support Node,3G-Gateway GPRS Support Node and IP server.
|Figure2. The tester offers realistic network simulation, and providesInternet connectivity with real data traffic flows.|
Network in a box
The Agilent8960 allows evaluation of IP throughput performance ofHSDPA/W-CDMA UE. The tester has a built-in HSDPA/ W-CDMA networksimulator and software verification tools, and is designed specificallyfor HSDPA/ W-CDMA UE developers who require performing radio/protocol/HW/SW design verification and integration, as well as performanceevaluation.
Figure 2 above shows how thetester can be used to replace the subsystems of a HSDPA/W-CDMA network.The tester offers realistic network simulation and provides Internetconnectivity with real data traffic flows.
Additional capability can also be obtained from extensive real-timeprotocol logging and analysis tools. Moreover, the tester can serve asa platform for comparing the throughput of different UEs by providing astable and reproducible network environment.
|Figure3. The UE initiates an HSDPA PSD connection with GPRS-attached andpacket data protocol activation processes.|
Agilent's HSDPA/W-CDMA network emulator has been selected for datathroughput experiments. The test set provides a complete, end-to-endpacket data service conforming to the GPRS Packet Data Service.
The user may connect the test set's LAN port to their computernetwork and make a complete connection from the UE through the test setto that network. The GPRS service configuration supported by the testset provides a number of different Radio Access Bearer configurations.The particular configurations studied here are 64k uplink (UL)/384kdownlink (DL) packet-switch data (PSD) mode, and 64k UL/3.6M DL PSDmode.
Figure 3 above shows theexchange of protocol layer messages between the UE and the wirelesstester. The messages indicate that the cellphone initiates to make apacket data connection with GPRS-attached and packet data protocolactivation processes. In the first test, a PC is connected to a UE anduses it as a wireless modem to download files of different sizes fromthe server via the test set.
Figure 4 below plots thetime taken to download vs. the file size for two different UEs. It canbe observed at DL data rates of 384Kbps and 7.2Mbps that both UEs havesimilar performance and that UE1 has slightly outperformed UE2.
|Figure4. At DL data rates of 384Kbps and 7.2Mbps, UE1 has slightlyoutperformed UE2.|
Moreover, the time taken increases linearly with the file size. Theraw data rates, computed from the inverse of the slopes of the lines,are approximately 195Kbps and 5,000Kbps for the 384Kbps DL and 7.2MbpsDL, respectively. The results show that when the DL speed increases bysix times from 384Kbps to 7.2Mbps, the actual increase in raw data rateis increased more than 25 times.
In the second test, the UEs operate alone and download files ofdifferent sizes. Figure 5 below plots the time taken to download against the file size for the two UEs.The performances of the UEs are similar to those in the previous case,except in one scenario.
|Figure5. At a DL data rate of 7.2Mbps, UE2 requires a substantially longertime to download files, giving a much lower raw data rate ofapproximately 1Mbps.|
At a DL data rate of 7.2Mbps, UE2 requires a substantially longertime to download files, giving a much lower raw data rate ofapproximately 1Mbps – possible causes of which are lower processingpower and smaller mobile memory buffer in UE2.
Thus, we can conclude that the design of UE1 is better than that ofUE2 in downloading data. Comparing modem and non-modem conditions andplot data analysis, it will be much easier to familiar with mobileperformance.
The Agilent 8960 tester provides a stable and reproducibleenvironment for engineers to evaluate the performance of UEs. Thesimulator has been implemented successfully and the IP data throughputof two different brands of UEs have been measured under two differentscenarios. With the 8960 tester, engineers can measure and optimize theperformance of the UE more easily.
Michael Leung is Applications Program Manager at AgilentTechnologies Inc.