Journal of Medical Engineering & Technology, Vol. 34, No. 3, April 2010, 172–177

Innovation Wireless communication interface for EEG/PSG Holter monitor ERNESTO VELARDE REYES*{, FRANCISCO MARANTE RIZO{, BELKIS MORGALO SANTOS{, JORGE GARROTE JORGE{ and FRANCISCO MARTIN GONZALEZ{ {Cuban Neuroscience Center, Playa, Havana, Cuba {Jose´ A. Echeverrı´ a Superior Polytechnic Institute, Marianao, Havana, Cuba (Received 17 June 2009; revised 28 October 2009; accepted 10 November 2009)

Communication interfaces for medical devices are normally wired. For long-term monitoring applications, wired devices limit patient mobility. In this paper a wireless communication interface for an EEG/PSG Holter monitor is presented. Selection of broadcasting band, communication standard, available hardware, and connection algorithm to use are discussed before making a choice. Results of experimental tests carried out on the prototype demonstrate the functionality of the implemented interface. Keywords: Wireless communication interface; Holter monitor; EEG; PSG .

1. Introduction Diagnosis of epilepsy and other neurological disorders can require lengthy recordings (up to three continuous days). This is also the case of polysomnography (PSG) studies, where the patient should be monitored for 8–10 hours. These kinds of application usually employ EEG/PSG Holter monitors due to the freedom of movement they offer to the patient. Holter monitors are portable devices that can record medical data in an ambulatory way. Several commercially available Holter monitors use wireless communication interfaces to transmit data to a PC (personal computer) during the recording. Some of them are: . Advanced Brain Monitoring developed a wireless sensor system based on IEEE 802.15.1 (known as Bluetooth) [1]. . Mind Media BV developed NeXus-4, a 4 channel physiological signal monitoring system, based on IEEE 802.15.1 [2]. . Nihon Kohden developed a 64-channel EEG/PSG system named AirEEG based on IEEE 802.11b (known as WLAN or WiFi) [3].

Compumedics developed a 32-channel EEG/PSG system named Siesta based on IEEE 802.11b [4].

Although there are some Holter monitors with wireless communication interfaces, we could not find any study or application note explaining their design and functioning. Therefore, the aim of this paper is to describe the design and implementation of a wireless communication interface for an EEG/PSG Holter monitor.

2. Implementing the wireless communication interface 2.1. Selecting the broadcasting band and characterization of the communication channel The selection of the broadcasting band to use must take into consideration the ability of the equipment to operate in different countries. Many commercially available Holter monitors use the 2400–2483.5 MHz radio band, one of the most popular frequencies of the industrial, scientific, and medical (ISM) license-free band. This band is used by numerous devices such as microwave ovens, wireless phones, RFID units and WLAN, causing a potential source of strong interference between devices within a

*Corresponding author. Email: [email protected] Journal of Medical Engineering & Technology ISSN 0309-1902 print/ISSN 1464-522X online ª 2010 Informa UK Ltd. http://www.informaworld.com/journals DOI: 10.3109/03091900903480747

173

Wireless communication interface for EEG/PSG Holter monitor

domestic environment. However, taking into account the small size required for the antenna, its good propagation characteristics and the existence of cheap transceivers and certified modules for it, we decided to choose this same band. Holter monitors are normally used indoors; therefore, we have to think about propagation of electromagnetic waves inside buildings. For this type of communication channel the main issues to study are interference coming from coexisting devices and the fading rate [5]. As we said above, many devices transmit in the broadcasting band. In addition, in the hospital environment there is interference caused by the action of medical equipment, such as electrical scalpels, microwave physiotherapy equipment and others [6]. Related to the fading rate, inside buildings there are a lot of obstacles that can attenuate electromagnetic waves. Walls, corridors, doors, windows and furniture cause penetration losses depending on the materials they are made up of. Moreover, inside buildings diffraction and multiple reflection propagation can occur. These phenomena are significant in 2400–2483.5 MHz bands and can cause fading. In order to mitigate the effect of interference, diffraction and multiple reflection propagation, it is necessary to use a wireless communication interface with a modulation technique that is effective against these phenomena. 2.2. Bit rate demands of EEG/PSG Holter monitor In EEG and PSG recording a minimum sampling rate of 200 Hz is recommended [7]. Our Holter monitor is a 40-channel system (figure 1) using a frequency of 200 Hz as sampling rate and a 24 bits encoded. Therefore, it needs a bit rate of 192 kbps for data transmission.

2.3. Selection of wireless communication standard First, we considered the possibility of a proprietary solution for implementation of the wireless communication interface, as there are a lot of commercially available 2.4 GHz transceiver chips (e.g. TI and Freescale). This solution was rejected due to the difficulty of communication with other standardized devices. We also studied other options, such as the well-known standards IEEE 802.15.4 (known as Zigbee), IEEE 802.15.1 (known as Bluetooth), and IEEE 802.11 (known as WLAN). Zigbee is a wireless technology that can reach an overthe-air bit rate of up to 250 kbps, with 128 kbps for the information [8,9] and the remainder for the protocol. Due to its throughput, this standard cannot be used in a wireless communication interface for our Holter monitor. Bluetooth and WLAN use different spread spectrum modulation techniques. These techniques are inherently resistant to interference, diffraction and multiple reflection propagation, and so are suitable modulation techniques for our wireless communication interface. These standards have different characteristics according to power consumption, throughput, etc (table 1). Bluetooth could have been selected as our wireless communication standard due to its throughput, but it does not have the advantages of the WLAN option. Due to its larger throughput and the number of devices that can be attached to its networks, WLAN is the protocol commonly used in PC-based wireless networks. Some hospitals use this telecommunication technology in automation of processes. Doctors can immediately access medical records and patients’ special medication [10,11]. If our wireless communication interface supports WLAN, we can guarantee monitoring of patients at all times and in all places in the hospital, by adapting into existing WLANs or installing new ones. This makes WLAN a very convenient option. Another advantage of WLAN is the ability to connect to a LAN via access points (AP), allowing the development of future applications of telemedicine. There are also various techniques to reduce the negative effects of interference [12–14]. Therefore WLAN is the wireless communication standard selected for our Holter monitor. WLAN or IEEE

Table 1. Some differences between Bluetooth and WLAN [19,20].

Power consumption Modulation technique Throughput Max number of devices in the basic cell

Figure 1. Diagram of a 40-channel EEG/PSG Holter monitor.

Bluetooth

WLAN

Low (1–35 mA) FHSS 3 Mbps Eight active devices; 255 in park mode

High (100–400 mA) DSSS, CCK, OFDM 54 Mbps Unlimited in ad-hoc networks; up to 2007 devices in infrastructured networks

174

E. V. Reyes et al.

802.11 is a family of specifications. Among them, IEEE 802.11b and IEEE 802.11g are the most popular and our wireless communication interface must support them. 2.4. Selection of an IEEE 802.11 module for the wireless communication interface In the short term, among the options to develop a module using integrated circuits and to acquire a certified module, the most economic one is the second. The selected characteristics of the IEEE 802.11 module were IEEE 802.11 b/g accomplishment, current consumption, host interface, embedded antenna and the implementation of a TCP/IP stack. Table 2 shows some of the revised modules. The OWSPA311g module from ConnectBlue was selected because it was available for the validation tests. It consumes low power, has a TCP/IP stack, and an easy AT command set. Moreover it is certified with R&TTE (Europe), FCC 15.247 (USA), and IC (Canada) directives [15]. However, for our application either of the modules could have been selected.

transport layer protocol (figure 2). This algorithm must ensure connectivity and a good performance in any network topology. Also, it must be simple, with low program processing times and therefore easy to implement by many programmable devices. The proposed algorithm has three stages: annunciation, acceptation and TCP connection establishment. In the annunciation stage, the EEG/PSG Holter monitor (the Client from now on) must begin the communication as soon as it has been authenticated and associated in a WLAN. Communication is initiated by the transmission of a command sequence that travels within UDP datagrams addressed to the broadcast address and to seven consecutive registered ports. The reason for using seven registered ports is that it is likely that one or more of them is being used by an application. Commands consist of annunciation indicatives that will reach all hosts inside the same physics network. Annunciation indicatives are formed by an ASCII character sequence integrated by: . .

3. Implementation of the connection algorithm EEG/PSG Holter monitors must be authenticated and associated in a WLAN for transmitting data using an IEEE 802.11 module. The WLAN can be an infrastructure or adhoc network. In the first case authentication and association processes are executed before an AP, in the second one the processes are executed before another module or similar equipment. A successful data interchange must be authenticated and associated, but it is also necessary to know the destination and source IP addresses and to possess a suitable TCP/IP stack. Source and destination devices must implement a network layer protocol for IP address dynamic assignation (by a network). The OWSPA311g module supports Dynamic Host Configuration Protocol (DHCP). 3.1. Connection algorithm The EEG/PSG Holter monitor must support a connection algorithm that resolves a destination IP address (in this case a PC) and enables medical data transmission through a

Table 2. Some IEEE 802.11 b/g modules [15,21–24]. Module OWSPA311g WISMC01BI-01 Radion WLAN6102EB WiPort

Current (Rx, Tx)

Interface

Antenna

TCP/IP stack

165 mA 250 mA 170 mA 112–223 mA 394–400 mA

UART UART SPI SPI UART

Yes Yes Yes No No

Yes Yes Yes No Yes

. .

.

.

a fixed-length identification string (e.g. Neuro); one character corresponding to the hexadecimal number that will indicate the number of characters in the Client IP address; the Client IP address; one divider character (e.g. /) that will separate the IP address of the registered port opened by the Client for the communication by means of UDP; one character that will be the hexadecimal number corresponding to the number of characters of the registered port opened by the Client; and the registered port opened by the Client.

An example is the string NeuroB192.168.0.2/44001, where the characters corresponding to hexadecimal numbers are underlined. During the annunciation stage, the indicatives transmitted by the Client arrive to all hosts connected to the same physical network, but are processed only by applications that opened the same registered ports as the UDP datagram destination. The application running on a PC (the Server from now on) opens these ports and receives the annunciation indicatives. Only the Server knows how to extract the IP address and the registered port number from the datagrams. As soon as the Server obtains that information, it permits data transmission to the Client. This new stage is known as acceptation and it is executed by Server in automatic or manual (by the user) form. The Server can simultaneously allow various Clients to pass to the next stage. Acceptation is accomplished by sending commands to the Client IP address. These commands travel in UDP datagrams and consist of acceptation indicatives that are integrated by an ASCII character sequence consisting of a fixed-length identification string (in this case NeuroACK)

Wireless communication interface for EEG/PSG Holter monitor

175

Figure 2. Diagram of connection algorithm between the Client and the Server.

and a character sequence similar to that of the annunciation indicative. The only difference in this case is that the Server indicates in its command the TCP port opened for the next connection. An example would be the string NeuroACKB192.168.0.1/45000, where the characters corresponding to hexadecimal numbers are underlined. The indicatives transmitted during annunciation and acceptation stages can be encrypted using an appropriate algorithm. After the acceptation stage, where the Client receives from the Server its IP address and port for establishing a TCP connection, the next stage is TCP connection establishment. In this stage, the Server opens the TCP registered port indicated in the acceptation command. Then it starts a passive opening and waits for TCP connection establishment (this process is described in the three-way handshake process of the TCP protocol [16]). After establishment, the Server begins data transfer, sending to the Client a TCP connection established indicator. This indicator consists of an ASCII character sequence (e.g. Go). As soon as the Client receives the TCP connection established indicator, it begins the data transmission to the Server. The Server transmits the indicator at every a fixed time interval (e.g. three 3 seconds). With the indicator, the Server transmits configuration commands to the Client. The Client replies to the indicator with a status indicator incorporating parameters such as sample frequency and active channels.

The proposed connection algorithm does not include the end of a TCP connection as the TCP protocol establishes. A connection is considered ended if the status and TCP connection established indicators do not arrive during a predefined time. For the establishment of new TCP connection, the Client and Server must reboot the connection algorithm from its first stage. 4. Validation of wireless communication interface For validating the wireless communication interface we built a prototype integrated by an OWSPA311g module and a microcontroller. The broadcasting power selected for the module was þ17 dBm and the microcontroller used was MSP430f1611 from Texas Instruments (Austin, Texas, USA). The MSP430f1611 was used for executing the tasks assigned to the Client in the connection algorithm. For programming the microcontroller’s firmware, IAR C/Cþþ Compiler for MSP430 v3.41 was used. All codes were implemented in C. A PC was used as the Server. To execute the Server’s tasks, it was implemented as a Virtual Instrument using Labview v8.2.1. The DWL-G132 D-Link adapter was used as wireless communication interface for PC. It was set up for broadcast with a power of þ17 dBm. Two experiments were selected for modelling the two principal scenarios that could face a wireless EEG/PSG Holter monitor inside a hospital. They were:

176

E. V. Reyes et al.

. Experiment 1 consisted of broadcasting data inside a room from a fixed position using multipath trajectories. This situation modelled a hospital room scenario due to the geometry of the room and the materials of the walls and furniture [17]. In the experiment the packet error rate (PER) is measured and the faults in the connection algorithm were counted. . Experiment 2 consisted of broadcasting data inside a corridor using line of sight (LOS) trajectories. The corridor selected is similar to other modern hospital corridors, with similar dimensions and materials [17]. In the experiment the PER is measured and the faults in the connection algorithm are counted. PER measurement consists of deducting the packet amount received by Server from the packet amount dispatched to the wireless module. To achieve this, the UARTs flow control was deactivated from the OWSPA311g module and the MSP430f1611 microcontroller. In the broadcastings 120-byte packets (ASCII characters) were sent every 5 ms, from MSP430f1611 to the OWSPA311g module through one of its UART interfaces. The UART was set up with a baud rate of 230 400 bps. Each packet was assembled by a header (sequence number), a separator (comma), and a character sequence. The OWSPA311g module has a broadcasting buffer capable of storing up to 2 kB of data (almost 17 120byte packets). Over PER measures influence the packet repetitions predicted in TCP protocol and the packet repetitions predicted by MAC Layer of IEEE 802.11 standard. In this paper, PER measurement only quantifies the packets lost in the entire system (Client, radio channel, Server) and not the quantity of retransmitted packets that arrive at their destination without error. It is very important to mention that PER measures can vary according the geometry of the room and materials or locations of walls and furniture. There are different models that can predict these variations with precision [18]. However, the selected experiments are sufficient to validate our wireless communication interface.

4.2. Experiment 2 This experiment was carried out in a 100-m long corridor, built with reinforced concrete and aluminum panels at distances ranging from 3 and 7 m. The Client and Server were placed in the centre of the corridor, at 1 m from the floor and at distances corresponding to multiples of 10 m up to the distance of 90 m (figure 4).

Figure 3. Diagram of the electronic design department of the Cuban Neuroscience Center (all distances are in metres).

4.1. Experiment 1 The Client and the Server were located in fixed positions in an office department similar to some hospital rooms. Both were positioned at a distance of 1 m from the floor. In the broadcastings, obstacle configuration inside the office guaranteed the multipath trajectories (figure 3). The Client sent 720 000 packets in three repetitions without any packet loss (PER ¼ 0%). In all repetitions, the connection algorithm ensured connection between the Client and the Server.

Figure 4. The 100-m long corridor. The Client is the car and the Server is the PC.

Wireless communication interface for EEG/PSG Holter monitor

The PER was measured three times per position without any packet loss and in each one were transmitted 65 535 packets. As in Experiment 1, the connection algorithm ensured connection between the Client and the Server in all repetitions. 4.3. Considerations about the experimental results The experimental results show the following: . The proposed connection algorithm guarantees the wireless communication interface connection. . The wireless communication interface works satisfactorily in multi-path and LOS trajectories. 5. Conclusions and future work The main result of this paper is the design and implementation of a wireless communication interface for an EEG/ PSG Holter monitor. As a result of the experiments we recommend that: (1)

(2)

an EEG/PSG Holter monitor prototype is developed that includes the interface implemented for validation in hospital environment; and the packet latency is measured in different environments, including LAN (local area network) networks with different configurations and traffic levels.

Acknowledgements Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper. References [1] Advanced Brain Monitoring, 2005, EEG Technology Review: Wireless Sensor Headset, B-Alert Software and Alertness and Memory Profiler (AMP) (Carlsbad, CA: Advanced Brain Monitoring). [2] Mind Media, 2007, NeXus-4 (Mind Media). Available online at: http// www.mindmedia.info [3] Nihon Kohden, 2007, AirEEG (Nihon Kohden).

177

[4] Compumedics, 2007, Siesta: Product Detail (Compumedics). [5] Parsons, J.D., 2000, The Mobile Radio Propagation Channel, 2nd edn (Chichester, UK: John Wiley & Sons). [6] Boyle, J., 2006, Wireless technologies and patient safety in hospitals. Telemedicine and e-Health, 12, pp. 373–382. [7] American Clinical Neurophysiology, 1994, Guideline 14: Guidelines for recording clinical EEG on digital media. Journal of Clinical Neurophysiology, 11, 114–115. [8] Webb, W., 2004, Low-rate networks take control: rest assured. Available online at: http://www.edn.com/contents/images/443378.pdf (accessed on December 11, 2009). [9] Jennic, 2006, Calculating 802.15.4 data rates. Application Note JNAN-1035. [10] Goldman, J., 2008, Wireless hospital: Orlando regional healthcare. Wi-Fi Planet. [11] Romero, P., 2006, Hospitales que son WiFi. El Mundo. [12] Gummadi, R., Wetherall, D., Greenstein, B. and Seshan, S., 2007, Understanding and mitigating the impact of RF interference on 802.11 Networks. Computer Communication Review. [13] Golmie, N., 2004, Interference in the 2.4 GHz ISM band: challenges and solutions. National Institute of Standards and Technology. [14] Chevrollier, N., Montavont, N. and Golmie, N., 2005, Handovers and interference mitigation in healthcare environments. MILCOM, 2005, Vol 1, pp. 74–80. [15] ConnectBlue, 2007, OWSPA311g i/x (TM) CB-0908. Electrical & Mechanical Datasheet. [16] Comer, D.E., 2000, Internetworking With TCP/IP. Vol I: Principles, Protocols, and Architecture, 4th edn (New Jersey: Prentice Hall). [17] Scha¨fer, T.M., Maurer, J., von Hagen J., and Wiesbeck, W., 2005, Experimental characterization of radio wave propagation in hospitals. IEEE Transactions on Electromagnetic Compatibility, 47, 304–311. [18] Wo¨lfle, G., Wertz, P. and Landstorfer, F.M., 1999, Performance, accuracy and generalization capability of indoor propagation models in different types of buildings. Paper presented at Proceedings of 10th IEEE International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC ’99), Osaka, Japan, September 1999. [19] Lee, J.-S., Su, Y.-W. and Shen, C.-C., 2007, A comparative study of wireless protocols: Bluetooth, UWB, ZigBee, and Wi-Fi. Paper presented at The 33rd Annual Conference of the IEEE Industrial Electronics Society (IECON), Taipei, Taiwan, November 5–8, 2007, pp. 46–51. [20] Ferro, E. and Potortı´ , F., 2005, Bluetooth and Wi-Fi wireless protocols: a survey and a comparison. IEEE Wireless Communications, 12, pp. 12–26. [21] Ezurio, 2007, Product Brief: WISMC01BI-01. Intelligent Wireless LAN Module-802.11b/g-UW (Ezurio). [22] Data Hunter, 2008, DS-Radion OEM Summary Sheets (Data Hunter). [23] SyChip, 2005, WLAN6102EB 802.11g Product Brief (SyChip). [24] LANTRONIX, 2006, WiPort PB (LANTRONIX).

Copyright of Journal of Medical Engineering & Technology is the property of Taylor & Francis Ltd and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.