Problem Solving of Low Data Throughput on Mobile Devices by Artefacts Prebuffering
© Ondrej Krejcar. 2009
Received: 29 March 2009
Accepted: 11 November 2009
Published: 22 December 2009
The paper deals with a problem of low data throughput on wirelessly connected mobile devices and a possibility to solve this problem by prebuffering of selected artefacts. The basics are in determining the problem parts of a mobile device and solve the problem by a model of data prebuffering-based system enhancement for locating and tracking users inside the buildings. The framework uses a WiFi network infrastructure to allow the mobile device determine its indoor position. User location is used for data prebuffering and for pushing information from a server to PDAs. All server data are saved as artefacts with its indoor position information. Accessing prebuffered data on a mobile device can significantly improve a response time needed to view large multimedia data. The solution was tested on a facility management information system built on purpose with a testing collection of about hundred large size artefacts.
The mobile wireless devices (laptops, PDA devices, Smart phones, etc.) are commonly used with internet connection which is available almost everywhere and anytime these days. The connection speed of the two most common standards GPRS and WiFi varies from hundreds of kilobits to several megabits per second. In case of corporate information systems or some other types of facility management, zoological or botanical gardens, libraries, or museums information systems, the WiFi infrastructure network is often used to connect mobile device clients to a server. Unfortunately, the theoretical maximum connection speed is only achievable on laptops where high-performance components are used (in comparison to mobile devices). Other mobile devices like family PDAs or Smart phones have low-performance components due to a very limited space. The limited connection speed represents a problem for online systems using large artefacts data files. It is not possible to preload these artefacts before the mobile device is used in remote access state. This problem was found as a very important point. The rest of this paper specifies the problem and suggests a possible solution.
The goal is to complete the data networking capabilities of RF wireless LANs  with accurate user location and tracking capabilities for user needed data prebuffering. This property is used as an information base for an extension of existing information systems or to create a special new one.
An information about location is used to determine both an actual and future position of a user . A number of experiments with the information system have been performed and their results suggest that determination of the location should be focused on. The following sections describe also the conceptual and technical details about Predictive Data Push Technology Framework (PDPT).
2. The Problem of Low Data Throughput
Why can we not use the classical model of user's requests and server's response for large data files? It is because some large amounts of data (artefacts) are impossible to download to PDA device and to be displayed in relatively short time.
an Ethernet network (Server to WiFi AP),
WiFi Access Points,
wireless communication between WiFi AP and WiFi antenna of the PDA device,
a PDA WiFi antenna,
a PDA WiFi adapter,
a PDA CPU,
a PDA memory,
a graphical unit—display data to user.
For large amount data artefacts, the most important parts are those listed under 5 to 8. The first and the second components have a relatively high throughput when compared to PDA device components and do not require testing. The Wireless communication between WiFi AP and PDA WiFi antenna cannot be improved.
The fourth component is a WiFi antenna of the PDA. This antenna has a standard of 2 dB gain and cannot be improved due to the absence of a connector to an external antenna. It is thus not necessary to test the antenna.
A number of tests with several types of PDA devices (HTC Athena, HTC Universal, HTC Blueangel and HTC Roadster) have been undertaken. These PDA devices were connected through CISCO WiFi APs (Signal Strength quality 80%) to an FTP server. The FTP server holds 3 types of large artefacts (files) which were downloaded to the internal PDA memory.
Data transfer tests—PDA is connected through WiFi infrastructure—use of internal FLASH ROM memory.
Data size (MB)
Transfer Speed (kB/s)
In theory, it is possible to find the worst component of all components and try to improve it. With this idea in mind, components 5 to 8 have been tested. It is not possible to test the WiFi adapter directly. The only possible way is to update a driver. The hardware of the WiFi adapter was found as identical in most cases of classical PDA devices from HP, HTC, or Acer companies. State-of-the-art drivers were found for all test devices showing us exactly the same transfer rates. Unfortunately, no easier solution to improve this component is known at this stage.
A CPU component is of course one of the most important parts at all. The power of the CPU can be easily benchmarked. All tested devices have an Intel XScale CPU. The Blueangel is equipped with PXA 263 with 400 MHz, Universal has 520 MHz PXA 270, and Athena and Roadster have 624 MHz PXA 270. Let us try to compare the best devices—Athena and Roadster with same CPU unit. A single CPU has entirely different transfer rates. That is why the speed of CPU is not so important on this occasion.
The PDA memory is a very important part on the other side. There is a large room for improvement of every PDA because of an SD Card slot which is present in most cases. The test was performed with standard SD Cards (Table 3).
Data transfer tests—SPB Benchmark software—Internal flash memory of PDA devices.
The best tested SD Card (Table 3) is a Kingston 2 GB 120 (120 means the card is 120 times faster than standard single speed CD ROM (150 kB/s)). When compared to an internal flash ROM, the writing speed of internal flash ROM is lower the reading speed remains 2 or 3 times higher.
The problem of low data throughput begins exactly in writing speed of internal flash. All data files are first transferred to cache (use of Microsoft explorer or Opera browser) which is relatively slow. The use of an SD card could speed up the transfer.
The other problem particularly for large data artefacts is the free internal memory. All PDA devices have a very limited space and usually operate with a memory allowing for 20 to 50 MB of free space including Athena device. Therefore the use of SD Cards would be a solution.
2.1. Maximum Application Response Time
Nielsen  specified the time delay of application response to user request to 10 seconds . "During this time the user was focused on the application and was willing to wait for an answer.'' Nielsen's book  was published in 1994, but it is a basic literature for this phenomenon. Another interesting source  suggests that "decreases in performance and behavioural intentions begin to flatten when the delays extend to 4 seconds or longer, and attitudes flatten when the delays extend to 8 seconds or longer.'' Based on these sources the maximum application response time is set to 10 seconds.
During the set maximum response time, the requested data must be downloaded and showed to user on display (in case of remote request to server's data). The time period of 10 seconds is used to calculate the maximum possible data size of a file transferred from server to client (during this period). To achieve the best transfer speed 160 kB/s, the calculated file size is 1600 kB.
The next step is to define an average artefact size. The network architecture building plan is currently used as a sample database, which contains 100 files of an average size of 470 kB. During the 10-second period, the client application can download 2 to 3 files (depending on the actual connection capabilities).
The second problem is the extremely long delay in displaying files in certain original file types (e.g., AutoCAD in case of vector graphic or MS Office in general cases). An AutoCAD file type is used in most cases of facility management of modern building . In such cases the mobile user needs to view a selected building scheme (building area plan, gas line plan, etc.) immediately.
Data files displaying tests for 500 kB files (13 iterations).
Data file displaying time (s)
Application starting times for selected data types (13 iterations).
Application Start Speed (s)
MS Internet Explorer
PPC CAD Viewer
MS Word Mobile
Unfortunately, the displaying time delays of files with nonbasic types are unacceptable. A basic data format must be used to display files by PDA natively (BMP, JPG, GIF) without any additional striking time consumption. The solution is a conversion from any format to these native formats (for PDA devices). In case of sound and video formats, it can also be recommended to use basic data format (wav, mp3, wmv, and mpg). In case of sample database, the display time of artefact is only about a half second per 500 kB artefact. This short time delay is not considered to 10 seconds response limitation. If other file types are used, the delay for presentation of file must be included.
The end result of several real tests and subsequent calculations give a definition of artefact size as an average value of 500 kB. The buffer size may differ from 50 to 100 MB in case of 100 to 200 artefacts.
In order to provide the reader with more information, the next chapter describes how a position can be obtained from wireless networks background.
3. Predictive Data Push Technology Framework
In most cases the low software level cache is used  or the residing of chips on system desk is recommended  to improve the performance of a system when operating with multimedia content. Such techniques are not allowed on existing mobile device where the operation system exists. Only a software solution added on top of the OS can achieve the objective.
A combination of a predicted user position with prebuffering of data associated with physical locations bears many advantages in increased throughput of mobile devices. An interesting solution (Microsoft US patent ) in this field needs to know all information (AP location, SS power, etc.) of all wireless base stations in mobile device before the localization process can be started (see the Location Manager module ). Moreover, the Moving Direction Estimator module is also situated in a mobile device application. These two facts present limitations to changing wireless base stations structure or to computing power consumption. Another solution (HP US patent ) represents a similar concept. A Location Determination and Path Guide modules are situated in mobile device side too.
The created PDPT Framework is based on a model of location-aware enhancement. This concept enables to increase the real dataflow from wireless AP (server side) to PDA (client side). The fact that the throughput (Table 1) is low on wireless connected mobile devices is very important with regards to the idea of using a prebuffered data for increasing transfer speed through WiFi connection on PDA mobile devices.
The general principle of localization states that if a WiFi-enabled mobile device is close to such a stationary device—base station, it may "ask'' the provider's location position by setting up a WiFi connection. If the mobile device knows the position of the stationary device, it also knows that its own position is within a 100-meter range of this location provider. The location accuracy can be improved by triangulation of two or several visible WiFi APs [12, 13]. The PDA client will support the application in automatically retrieving location information from nearby location providers, and in interacting with the server. Naturally, this principle can be applied to other wireless technologies. The application is now implemented in C# using the MS Visual Studio .NET 2005 with .NET compact framework and a special OpenNETCF library enhancement. The information about the basic concept and technologies of user localization can be found in .
The current and predicted user positions are used for the PDPT framework to make decisions as to which data artefacts are needed in the PDA memory. The data prebuffering increases the primary dataflow from WiFi AP (server side) to PDA (client side). These techniques form the basis of the predictive data push technology (PDPT).
PDPTs push the data from an information server to the client's PDA on the basis of the user's location and user's future predicted location. The prebuffered data will be helpful when the user comes to the location which was predicted by PDPT Framework. The benefit of the PDPT consists in the reduction of time needed to display a desired information requested by a user command on the PDA. This delay may vary from a few seconds to a number of minutes. It depends on two aspects.
The first aspect is the quality of wireless WiFi connection used by the client PDA. A theoretical speed of WiFi connection is maximum 687 kB/s. However, the test suggests a speed of only 43–160 kB/s (depending on file size and PDA device) (Table 1).
The second aspect is the size of copied data. The current application records just one set of WiFi signal strength (SS) values at a time (by Locator unit in PDPT Client). From this set of values the actual user position is determined by the PDPT Server side. PDPT Core responds to a location change by selecting the artefact to load to PDPT Client buffer. The data transfer speed is to a large extent influenced by the size of these artefacts. For larger artefact size the speed decreases.
3.1. PDPT Framework Data Artifact Manager
The PDPT Server SQL database manages the information (e.g., data about Ethernet hardware such as Ethernet switch UTP socket, CAT5 cable lead, etc.) in the context of their location in building environment. This contextual information is the same as location information about user track. The PDPT Core controls data, which are copied from the server to the PDA client by context information (position info). Each database artefacts must be saved in the database along the position information to which it belongs.
The Manager allows to the administrator to create a new artefacts from multimedia files (image, video, sound, etc.) and edit or delete the existing artefacts. The left side of the screen contains the text field of artefact metadata as a position in 3D space. This position is determined by the artefact size (in case of building plan) or by binding of the artefact to some part of a building in 3D space. It is possible to take the 3D axis from a building plan by a GIS software like Quantum GIS or by own implementation [15, 16]. The central part represents a multimedia file and the right side contains the buttons to create, edit, or delete the artefact. The lower part of the application screen shows the actual artefacts in WLA database located on MS SQL Server.
3.2. The PDPT Framework Design
The PDPT framework design is based on the most commonly used server-client architecture. To process data the server has an online connection to the information system. Technology data are continually saved to SQL Server database .
3.3. PDPT Client
The PDPT Client is a Windows Mobile 6.1-based application. The PDPT Client was developed for testing and tuning the PDPT Core. This client realizes a classic client to server side and an extension by PDPT and Locator module.
The info about nearby APs is sent to the PDPT Server which responds with a number of recognized APs in the database (Locator AP ret. Text window). In the presented case, the 7 APs are in user neighbourhood, but only 2 APs are recognized by the PDPT Server database (info about 2 APs is in WLA database). The scanning interval is set to 2 seconds and finally the text "PDPT Server localization OK" means that the user PDA was localized in an environment and that this position can be used by the PDPT Core to prebuffer the data to the client device.
The middle section of the PDPT tab (Figure 7) shows logging info about the prebuffering process. The right side shows the time of artefact loading (part time) and the full time of prebuffering.
3.4. PDPT Client—Microsoft SQL Server CE Database
Compact and shrink of DB file means two options for manual database compression. The time in millisecond is measured in a text box in between the two buttons. Both of these mechanisms are used in prebuffering cycles when the large artefact is deleted from database table to release space of deleted artefact. The database file has occupied space of deleted artefact by default, because the standard operation of delete order does not include this technique. This is due to recovery possibilities in Microsoft SQL Server CE databases.
For a mobile device to determine its own position, it must have a WiFi adapter still alive. This fact provides a limitation of using of mobile devices. The complex test with several types of batteries is described in .
A number of indoor experiments were achieved with the PDPT framework using the PDPT Client application. The main result of the use of the PDPT framework is a reduction of data transfer speed. The tests focused on the real use of the developed PDPT Framework and its main impact on increased data transfer.
A special Biotelemetry system for patient monitoring is under development at our department. In this complex system the wide network of remote sensors is used to collect data. This system proved to be a useful platform for prebuffering the large data-artefacts [19, 20]. Localization module of PDPT framework is suitable for home security system . For any kind of emergency cases, the special wireless network MANET can be suitable improvement of PDPT solution to avoid any problems in case the signal of preferred WiFi network is missing .
The problem of low transfer rates in mobile devices was presented. Some suggestions have been put forward (e.g., to use a high-performance SD Cards for large data amount to get a higher transfer rate). The low transfer rates problem was considered also in the context of a maximum response time for user requests.
The PDPT Framework was described as one of the possible solutions. The indoor location of a mobile user is obtained through an infrastructure of WiFi APs. This mechanism measures the quality of the link of nearby location provider APs to determine the actual user position. User location is used in the core of server application of the PDPT framework to data prebuffering and pushing information from the server to the user's PDA. Data prebuffering is the most important technique to reduce the time from a user request to system response. The experiments show that the location determination mechanism accurately and with sufficient quality determines the actual location of the user in most cases. Minor inaccuracies do not impact significantly on the PDPT Core decision making. The framework was evaluated in a real use experiment.
This research has been carried out under the financial support of the research grant "Centre for Applied Cybernetics,'' Ministry of Education of the Czech Republic under Project 1M0567.
- Evennou F, Marx F: Advanced integration of WiFi and inertial navigation systems for indoor mobile positioning. EURASIP Journal on Applied Signal Processing 2006, 2006:-11.Google Scholar
- Olivera VM, Plaza JMC, Serrano OS: WiFi localization methods for autonomous robots. Robotica 2006, 24(4):455-461. 10.1017/S0263574705002468View ArticleGoogle Scholar
- Kostuch A, Gierłowski K, Wozniak J: Performance analysis of multicast video streaming in IEEE 802.11 b/g/n testbed environment. In Wireless and Mobile Networking, IFIP Advances in Information and Communication Technology. Volume 308. Springer, New York, NY, USA; 2009:92-105. 10.1007/978-3-642-03841-9_9Google Scholar
- Nielsen J: Usability Engineering. Morgan Kaufmann, San Francisco, Calif, USA; 1994.Google Scholar
- Haklay M, Zafiri A: Usability engineering for GIS: learning from a screenshot. The Cartographic Journal 2008, 45(2):87-97. 10.1179/174327708X305085View ArticleGoogle Scholar
- Galletta DF, Henry RM, McCoy S, Polak P: When the wait isn't so bad: the interacting effects of website delay, familiarity, and breadth. Information Systems Research 2006, 17(1):20-37. 10.1287/isre.1050.0073View ArticleGoogle Scholar
- Krejcar O: User localization for intelligent crisis management. Proceedings of the 3rd IFIP Conference on Artificial Intelligence Applications and Innovation (AIAI '06), June 2006, Athens, Greece 221-227.View ArticleGoogle Scholar
- Asaduzzaman A, Mahgoub I, Sanigepalli P, Kalva H, Shankar R, Furht B: Cache optimization for mobile devices running multimedia applications. Proceedings of the IEEE 6th International Symposium on Multimedia Software Engineering (ISMSE '04), December 2004 499-505.View ArticleGoogle Scholar
- Rosner S, Mcclain M, Gershon E: System and method for improved memory performance in a mobile device. US patent, Spansion LLC, 20060095622, 2006Google Scholar
- Brasche GP, Fesl R, Manousek W, Salmre IW: Location-based caching for mobile devices. US patent, Microsoft Corporation, Redmond, Wash, USA, 20070219708, 2007Google Scholar
- Squibbs RF: Cache management in a mobile device. US patent, Hewlett-Packard Development Company, L.P., 20040030832, 2004Google Scholar
- Brida P, Duha J, Krasnovsky M: On the accuracy of weighted proximity based localization in wireless sensor networks. In Personal Wireless Communications, IFIP International Federation for Information Processing. Volume 245. Springer, New York, NY, USA; 2007:423-432.Google Scholar
- Brida P, Majer N, Duha J, Cepel P: A novel AoA positioning solution for wireless ad hoc networks based on six-port technology. In Wireless and Mobile Networking, IFIP International Federation for Information Processing. Volume 308. Springer, New York, NY, USA; 2009:208-219.Google Scholar
- Krejcar O, Cernohorsky J: Database prebuffering as a way to create a mobile control and information system with better response time. Proceedings of the 8th International Conference on Computational Science (ICCS '08), June 2008, Kraków, Poland, Lecture Notes in Computer Science 5101: 489-498.Google Scholar
- Horak J, Unucka J, Stromsky J, Marsik V, Orlik A: TRANSCAT DSS architecture and modelling services. Control and Cybernetics 2006, 35(1):47-71.MATHGoogle Scholar
- Horak J, Orlik A, Stromsky J: Web services for distributed and interoperable hydro-information systems. Hydrology and Earth System Sciences 2008, 12(2):635-644. 10.5194/hess-12-635-2008View ArticleGoogle Scholar
- Jewett M, Lasker S, Swigart S: SQL server everywhere: just another database? Developer focused from start to finish. DR DOBBS Journal 2006., 31(12):Google Scholar
- Krejcar O: PDPT framework—building information system with wireless connected mobile devices. Proceedings of the 3rd International Conference on Informatics in Control, Automation and Robotics (ICINCO '06), August 2006, Setubal, Portugal 162-167.Google Scholar
- Penhaker M, Cerny M, Martinak L, Spisak J, Valkova A: HomeCare—smart embedded biotelemetry system. Proceedings of the World Congress on Medical Physics and Biomedical Engineering (WC '06), August-September 2006, Seoul, South Korea 14: 711-714.Google Scholar
- Cerny M, Penhaker M: Biotelemetry. Proceedings of the 14th Nordic-Baltic Conference on Biomedical Engineering and Medical Physics (NBC '08), June 2008, Riga, Latvia 20: 405-408.Google Scholar
- Kasik V: FPGA based security system with remote control functions. Proceedings of the 5th IFAC Workshop on Programmable Devices and Systems (PDS '01), November 2001, Gliwice, Poland 277-280.Google Scholar
- Ramrekha TA, Politis Ch: An adaptive QoS routing solution for MANET based multimedia communications in emergency cases. Proceedings of the 1st ICST International Conference on Mobile Lightweight Wireless Systems (MOBILIGHT '09), May 2009, Athens, Greece, Lecture Notes of the Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 13: 74-84.View ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.