WIRELESS NETWORKS EFFICIENCY STUDY BASED ON

METEOROLOGICAL DATA MEASUREMENTS

P. Mariño, F. Machado, S. Otero

Electronic Technology Department. University of Vigo.

ETSEI (DTE) Lagoas Marcosende S/N, CP: 36200 Vigo, Spain

F.P. Fontán, C. Enjamio

Signal Theory and Communications Department. University of Vigo.

ETSET (TSC) Lagoas Marcosende S/N, CP: 36200 Vigo, Spain

Keywords:

Wireless Communications, Millimetre Wavelengths, Acquisition Systems, Performance Analisys.

Abstract:

Wireless access networks based in millimetre wavelength technologies are mainly impaired by rain. To eval-

uate the rain effects over a communication system, it is essential to know the temporal and spatial evolution

of rainfall rate. For this reason, it is necessary to develop an experimental network which provides the ade-

quate data to study, prevent and compensate the rain fade. In this paper, an experimental rain gauge network

is presented. This network comprises weather stations capable of measuring rainfall rate, temperature and

humidity. The paper ﬁrst describes the experimental network for automatic data acquisition as a system based

in a distributed process. The design of the experimental network is explained in detail and ﬁnally the interest

in millimetre wavelength applications is pointed out.

1 INTRODUCTION

The space-time variability of rain intensities at local

scale is an essential input for a number of studies,

including the planning and management of drainage

(Moore et al., 2000; Casale and Samuels, 1998) and

telecommunications networks (Redaño and Lorente,

1993; COST, Final Report: Action 255, 2002; COST,

Memorandum of Understanding: Action 280, 2001).

In the ﬁeld of meteorology and hydrology for ur-

ban areas, the requirements are for rainfall data with

very ﬁne time and space resolution. For this reason,

a number of experimental campaigns to obtain data

from weather radars and rain gauge networks have

been developed in the past. The analysis of these data

contributes to a better knowledge about the local dis-

tribution of precipitation (Moore et al., 2000; Redaño

and Lorente, 1993; Holland, 1967; Niemczynowicz,

1988). In some cases, it is possible to take advan-

tage of these networks and use their data for stud-

ies in the ﬁeld of telecommunications (Enjamio et al.,

2002a). In addition, the availability of similar net-

works placed in different climatic regions will permit

the cross comparison of the results about the spatial

structure of rain, and its inﬂuence in the performance

of radio telecommunication networks.

In the ﬁeld of telecommunications, recent studies

have shown the suitability of using weather radars to

predict the impairments caused by rain over satellite

or terrestrial links (Crane, 1980; Mannes et al., 2002).

Moreover, in Europe there are about 80 weather

radars used for meteorological purposes. Images from

weather radars have also been used for the characteri-

zation of the spatial properties of rain (Goldhirsh and

Musiani, 1992).

The rainfall rate can change dramatically along

space and time particularly during convective events

(Redaño and Lorente, 1993). Because of this, the rain

fade suffered by a terrestrial or satellite link can not be

exactly predicted from a point measurement of rain-

fall rate recorded, for example, at the receiver site.

Since the link has a certain path length that could be

partially or totally immersed within the rain, the atten-

uation series can be calculated more accurately if in-

formation about the spatial distribution of rainfall rate

along the path is provided, for instance, by a weather

radar or a rain gauge network.

For the experiment described in this paper both

data sources are considered, a weather radar which

belongs to the Spanish Meteorological Ofﬁce (SMO)

and a rain gauge network consisting in 20 tipping

bucket gauges located within the coverage of the

radar.

Current advances in electronic and communica-

tion technologies have permitted the development of

multifunctional sensor nodes with enhanced wireless

Machado F., Mariño P., P. Fontán F., Enjamio C. and Otero S. (2004).

WIRELESS NETWORKS EFFICIENCY STUDY BASED ON METEOROLOGICAL DATA MEASUREMENTS.

In Proceedings of the First International Conference on E-Business and Telecommunication Networks, pages 60-66

DOI: 10.5220/0001395600600066

 SciTePress

Figure 1: Weather station data Acquisition System (WAS)

structure for the rain gauge network.

communication capabilities (Akyildiz et al., 2002).

These nodes sense, process and transmit the data, en-

abling the development of automatic sensor networks,

which nodes are accessible from a wide area. These

communication capabilities allow the system data to

be remotely monitored and consulted. The implemen-

tation of a warehousing approach, allows data to be

stored in a centralized database system that is respon-

sible for query processing (Bonnet et al., 2000; Mar-

iño et al., 2003a). The stored data will be used in the

future to analyse the spatial and temporal variability

of the rainfall rate.

The paper describes ﬁrst the different elements em-

ployed in the experiment. These included (i) the me-

teorological sub-network, (ii) the propagation sub-

network and (iii) the data management. The paper

progresses towards the application of these data to

performance studies of radio networks operating at

millimetre wavelengths.

2 DATA ACQUISITION SYSTEM

The Weather station data Acquisition System (WAS)

structure for meteorological variable measurement is

based on the GSM mobile telephony network, for

communications between control station (CS) and

weather stations (WS), and the Internet for commu-

nications with the weather radar (WR) and the central

wireless node (CWN) (Fig. 1).

In this system the following physical and logical

elements have been designed:

• Twenty weather stations with all their sensors and

structural elements necessary for their installation

in ﬁeld.

• GSM, Internet and LAN networks access conﬁg-

uration, for data transmission and users queries

(Fig.1).

• Communication and data processing control station

(CS), that enables the information transfer between

weather stations and the database.

• Database (DB) for meterorological parameters

storage and analysis, and the corresponding inter-

faces.

• Data formats for meteorological information pro-

cessing, storage and modelling.

A second communication network has been su-

perimposed to the weather stations experimental net-

work, to study the precipitation impact on millime-

tre wavelength communication system. This second

network comprises the following physical and logical

elements:

• Three broadband radio links in a point-to-

multipoint (P-MP) conﬁguration at 24 GHz formed

by three secondary wireless nodes (SWN), and one

central wireless node (CWN) with Internet connec-

tion.

• Three WLAN based point-to-point links at 2.4 GHz

between the CWN and the SWNs.

• Communication and data processing control soft-

ware for the CS.

• Database for propagation parameters storage and

its corresponding interfaces.

• Data formats for radio network propagation infor-

mation processing, storage and broadband service

planning.

Each weather station comprises a GSM modem that

transmits the meteorological information to the CS

through a data call. In order to reduce costs, the CS

makes a call to all the WSs every 24 hours by means

of a polling procedure. During these calls the WSs

send all the information that has been stored during

that period. On the other hand, the collected data from

the WR (radar images) and the CWN (radio links

state) are sent through direct connections to Internet,

under CS request. Therefore the CS periodically exe-

cutes the reading data process and later database stor-

age of the received information, through an Ethernet

local area network.

The Territorial Meteorological Centre, which be-

longs to the Spanish Meteorological Ofﬁce (Fig. 2),

queries the database by Internet. Therefore, it is

possible to calibrate and compare the data from

the weather radar connected to the SMO, with the

weather station data from the experimental rain gauge

network. This allows to execute different applications

whose meteorological impact can be relevant (section

4).

2.1 Meteorological Data

The meteorological data sub-network consists on a C-

band weather radar and a rain gauge network com-

WIRELESS NETWORKS EFFICIENCY STUDY BASED ON METEOROLOGICAL DATA MEASUREMENTS

Figure 2: Weather station grid and weather radar location

map.

posed of 20 tipping-bucket gauges, located within the

radar coverage and in the surroundings of the city of

A Coruña (Northwest of Spain).

The weather radar belongs to the Spanish Meteoro-

logical Ofﬁce and it is located on a hill 600 m above

sea level. The radar can operate in two modes: (i)

normal mode which provides a horizontal resolution

of 2 Km and (ii) Doppler mode with 1 Km resolution.

It performs twenty down-up scans resulting in a to-

tal time interval between two successive volumetric

scans of 10 minutes. The reﬂectivity data recorded by

the radar are then sent to the Regional Meteorological

Ofﬁce through a radio link at 10 GHz.

The chosen topology for the rain gauge network

was a uniform grid of square cells. The cells have a di-

mension of 3x3 Km. Following this regular topology,

one weather station has been installed within each cell

to complete a total of 20 rain gauges (Fig. 2).

A weather station (WS) is the basic acquisition sys-

tem and carries out the data registration (measure-

ments and processing), and the communication with

the control station (CS). In this way, each WS com-

prises an automatic measurement unit with data trans-

fer capacity. Each WS is located within one of the

squares of the grid shown in ﬁgure 2.

The rainfall rate measurements are carried out by

an electronic tipping-bucket gauge with 0.1 mm tip

size. The rain drops are captured by a collector and

sent to a small bucket. When this bucket is full, it

tips out, closing an electronic circuit and sending an

Figure 3: WS data acquisition and communications system

assembled in the protection box.

impulse to the console.

The console is the system nucleus. It captures

the data from each sensor, automates the measure-

ments, synchronises the data and manages the com-

munications. The data transmission is carried out by

means of the data logger and the GSM modem con-

nected to the console. Next, the data captured by the

console is sent to the storage system where they are

saved. The communication process setting, through

the GSM modem connected to the data logger, per-

mits the control and programming of several tasks in

the console as well as the acquisition of the stored

data.

The data captured by the console are organised in

registers. The registers comprise the sensor outputs as

well as the time and date. These registers are then sent

to the storage system where they are saved for a future

access. The console is programmed for capturing and

storing the sensors information each minute. Due to

the limited capacity of the storage system integrated

in the WS, the data can only be stored during a day

(24 hours).

Figure 3 illustrates the console, the storage sys-

tem, the GSM modem and connections with the sen-

sors and the electrical supply. All these elements are

placed inside a box which protects them from the

weather conditions. This box as well as all the sensors

are ﬁxed to a metallic base located at the site (Fig. 4).

2.2 Propagation Data

The second part of the experiment comprises the

study of the impairments caused by rain, in the ra-

dio communications systems working in the millime-

tre wavelengths. For this reason a sub-network com-

prising three radio links at 24 GHz has been installed,

ICETE 2004 - WIRELESS COMMUNICATION SYSTEMS AND NETWORKS

Figure 4: Final assembly and WS installation.

within the area covered by the meteorological data

network (Fig. 2). The topology was chosen to be

point-to-multipoint (P-MP) or divergent in order to

take advantage of the spatial structure of rain and the

path length is about 4 Km. Consequently, the links re-

produce a network conﬁguration which beneﬁts from

angular diversity. With this particular conﬁguration,

it is possible to study the relationship or dependence

between the three attenuation series in terms of corre-

lation, angular-diversity improvement, etc. The ra-

dio links are unidirectional from the CWN to the

SWNs. The attenuation data registered at 24 GHz is

sent through a signalling channel at 2.4 GHz employ-

ing WLAN technology. The signalling channel is not

affected by rain and sends the attenuation registered

by the three links to the CWN, where it is again sent

to the CS through the Internet.

2.3 Global Data Management

The information obtained from the data acquisition

system (meteorological and propagation data) is col-

lected by the control station and stored in the database

for later process, analysis and query.

The CS requests and compiles the data from the

different elements of the developed system (Fig. 5) to

store them in the database. Thus the CS is provided

with a GSM modem to make the polling query of each

experimental WS in the network, and with an Inter-

net connection to check the radio link state from the

CWN, get the radar images from the WR and com-

municate with the DB.

Therefore the CS is a PC connected to GSM net-

Figure 5: WAS architecture and database interfaces.

work and Internet that executes the developed pro-

gram to perform its operations ﬂowchart. The ﬁgure

6 shows this ﬂowchart detailing the three basic tasks

of system data query: WSs, CWN and WR query.

Since all the measured data must have the same

time reference for its later process, the control station

obtains the system reference clock from a real time

network server by the NTP synchronization protocol

(Network Time Protocol). So after the data have been

obtained, a time synchronization test is veriﬁed for

the WSs, the CWN and the WR clocks, to determine

if the collected data can be considered valid. If this is

the case, the information is stored directly in the DB.

In the opposite situation, the problem is corrected (if

it is possible), it is notiﬁed by email and/or a SMS

message, and ﬁnally the data and the error informa-

tion are stored. In this way it is possible to known

exactly when and what type of errors took place and,

depending on this information, weather data can be

corrected.

The data from the weather stations network, the ra-

dio links and the WR are centralized in a relational

database. This DB presents one interface with the

CS, through which all the system information is in-

troduced, and three interfaces to access this informa-

tion: general data accesses, access to interesting data

to analyse the wireless services, and query of data to

obtain models (Fig. 5).

The interface between CS and DB is executed di-

rectly by means of calls, from the program running in

the CS, to the API functions of the database manage-

ment system (DBMS). This optimises the collected

data storage. However the queries do not have the

acquisition system restrictions (Mariño et al., 2003a)

and it is worth being generic interfaces. For this rea-

son this queries have been made by means of ODBC

(Open DataBase Connectivity).

The general data access will directly take place

through an Internet accessible web page. Whereas

for queries related to the analysis of wireless services

and models, the access is made through speciﬁc views

for each type of study (Mariño et al., 2003b). These

views (virtual cards) considerably facilitate the ser-

WIRELESS NETWORKS EFFICIENCY STUDY BASED ON METEOROLOGICAL DATA MEASUREMENTS

vices and models analysis.

query

WR data

storage

WR state

verification

(Synchronization

and failure)

Start

query

Weather registers

storage

WS state

verification

(Synchronization,

sampling and failure)

Alarms

activation

CWN

query

Radio link state

storage

Errors?

Resynchronization

and reconfiguration

CWN state

verification

(Synchronization

and failure)

Notification

(email, SMS)

Temporary

Storage

Permanent

Storage

YES

Figure 6: CS operations ﬂowchart.

3 APPLICATION TO

BROADBAND RADIO

TECHNOLOGIES IN

MILLIMETRE WAVELENGTHS

The suitability of employing weather radars for prop-

agation studies in the millimetre wavelengths was

already pointed in the introduction. Nevertheless,

weather radar accuracy is mainly limited by two fac-

tors. First, the transformation of radar reﬂectivity

(Z) into rainfall rate (R) involves the assumption of

a certain Raindrop Size Distribution (RSD). Although

there are some well-known experimental RSD such as

the Marshall-Palmer or the Joss distributions, it has

been found that the RSD changes with the type of

storm (Vilar et al., 1997). The usual process to obtain

the rainfall rate consists in assuming a certain RSD

and obtaining the relation between Z and R. This pro-

cess may not be correct since it is based on the RSD

assumption which can not validated in practice.

Because of the availability of a rainfall rate

database provided by the rain gauge network, it is pos-

sible to "calibrate" the radar images with real rainfall

rate data.

The second radar limitation that must be taken into

account is the spatial and temporal resolution. The

weather radar used in this research provides a hor-

izontal resolution of 2 Km when it is operating in

normal mode and 1 Km when it operates in Doppler

mode. The time interval between two successive radar

volume scans is 10 minutes. The resolution can be

improved if a rain gauge network complements the

radar images because (i) the sensors are easy to in-

stall which permits getting the desired spatial resolu-

tion and (ii) in contrast with the radar images, the rain

gauges provide instantaneous rainfall rate data.

Weather radar limitations can then be reduced if

data from a rain gauge network is used. In addition,

the radar provides coverage in places where the instal-

lation of a rain gauge network is not feasible.

Attending the reasons above, the design of a rain

gauge network has been carried out as it was ex-

plained in section 2. The cells size was chosen to be

3x3 Km because of the similarity with previous ex-

periments (Moore et al., 2000; Enjamio et al., 2002a).

Nevertheless, the rain gauges are not uniformly ar-

ranged due to the difﬁculties to ﬁnd the ideal sites,

and the distances between them are, in some cases,

smaller or bigger than 3 Km. In any case, the average

distance provides a ratio of at least one rain gauge ev-

ery two radar pixels. In addition, recent studies have

shown that this distance is enough to distinguish and

extract the structure and movement of the rain cells

(Enjamio et al., 2002b).

Some previous studies about the spatial distribution

of rainfall rate have illustrated the signiﬁcance of us-

ing spatial data for the analysis/prediction of the be-

haviour of the telecommunications systems. Figure 7

shows an illustrative example of a measured attenua-

tion series carried out with the ITALSAT satellite at

39.6 GHz and the series calculated using rainfall rate

data from a dense rain gauge network (Enjamio et al.,

2002a).

The characterisation of the spatial distribution of

rainfall rates can be obtained from the radar images,

once calibrated using the rain gauges information.

Radar images can then be applied to the simulation

and analysis of the performance of communication

systems, ﬁxed terrestrial or satellite, in the millimetre

wavelengths. This analysis will permit the design and

evaluation of numerous ways of optimising the com-

munications by adapting the transmission parameters

such as the transmitted power, coding or modulation.

The spatial and temporal characterisation of rain-

fall rates is also useful for the development of mi-

croscale/mesoscale precipitation models. These mod-

els can then be applied to a number of systems includ-

ing hydrology or communications.

The analysis of the spatial data provided by the

radar-gauges network will complement the avail-

able information obtained from similar experiments

ICETE 2004 - WIRELESS COMMUNICATION SYSTEMS AND NETWORKS

Figure 7: Comparison between measured and computed at-

tenuation for the same radio link. Barcelona ITALSAT. 41

elevation, 39.6 GHz.

(Moore et al., 2000; Enjamio et al., 2002a). This fact

will help to reduce the lack of results about the time-

space structure of rainfall rate at local scale.

4 CONCLUSIONS

The authors have designed a meteorological and prop-

agation data acquisition and storage system (WAS)

based on GSM network and Internet. This system

has been developed within the area covered by the

weather radar of the SMO, located in the neighbour-

hood of A Coruña (Spain). The WAS comprise a me-

teorological data sub-network formed, in addition to

the WR, by a grid of 20 weather stations within 3 Km

cells, and a propagation data sub-network formed by

three radio links at 24 GHz and the signalling channel

at 2.4 GHz (Fig. 2).

The aim of the WAS experiment is not exclusively

limited to the ﬁeld of telecommunications and the

evaluation of the impairments caused by rain within

broadband wireless networks, like the studied one in

this article (section 3). For this reason the extension

to the general meteorology scope is taken into ac-

count, providing a web service where meteorological

data are available to all the users, and national or in-

ternational agencies related to environmental studies

in meteorology, hydrology, communications, natural

disasters prevention, etc.

In the scope of wireless optical technologies appli-

cations, also it is interesting the evaluation of optics

based services in free space (FSO) (Dornan, 2002).

Also, in the radio communications scope, it is

predicted to extend the study to other radio net-

works services such as: satellite distribution net-

works, satellite data networks with small aperture an-

tennas (VSAT), stratospheric platforms (HAPs) (Har-

ris, 2000), aeronautical services, ﬂeets radio commu-

nication (TETRA), mobile networks without infras-

tructure (MANET), etc.

ACKNOWLEDGMENTS

This work has been sponsored by two R&D projects

from the following entities: Research General Di-

rectorate of the MCYT, Ref. TIC2001-3701-C02-

01, Central Government (Madrid, Spain); and Presi-

dency Department, Ref. PGIDT01TIC30301PR, Au-

tonomous Government (Galicia, Spain).

REFERENCES

Akyildiz, I., Su, W., Sankarasubramaniam, Y., and Cayirci,

E. (2002). Wireless sensor networks: a survey. Com-

puter Networks, 38(4):393–422.

Bonnet, P., Gehrke, J., and Seshadri, P. (2000). Querying

the physical world. IEEE Personal Communications,

pages 10–15.

Casale, R. and Samuels, P. (1998). Hydrological risks:

Analysis of recent results from ec research and tech-

nological development actions. Technical report, Eu-

ropean Commision. Environment and Climate Pro-

gramme.

COST, Final Report: Action 255 (2002). Radiowave propa-

gation modeling for new satcom services at Ku-Band

and above. Technical report, Commission of the Eu-

ropean Communities, Brussels.

COST, Memorandum of Understanding: Action 280

(03/21/2001). Propagation impairment mitigation for

millimetre wave radio systems. Technical report,

Commission of the European Communities, Brussels.

Avaliable online. http://www.cost280.rl.ac.uk.

Crane, R. (1980). Prediction of attenuation by rain. IEEE

Transaction on Communications, COM-28(9):1717–

1733.

Dornan, A. (2002). Wireless optics: ﬁber is cheap, but space

is free. Network Magazine, pages 28–32.

Enjamio, C., Vilar, E., and Fontan, F. (2002a). Rainfall rate

spatial distribution at local scale: Rain cell analysis in

the mediterranean region. In URSI Open Symposium

on Propagation and Remote Sensing, Garmisch, Ger-

many.

Enjamio, C., Vilar, E., and Fontán, F. (2002b). Consider-

ations on short distance diversity gain and dynamic

rain fade. In 20th AIAA International Communication

Satellite Systems Conference and Exhibit, Montreal,

Quebec, Canada.

Goldhirsh, J. and Musiani, B. (1992). Dimension statis-

tics of rain cell cores and associated rain rate isopleths

WIRELESS NETWORKS EFFICIENCY STUDY BASED ON METEOROLOGICAL DATA MEASUREMENTS

derived from radar measurements in the mid-atlantic

coast of the United States. IEEE Transactions on Geo-

science and Remote Sensing, 30(1):28–37.

Harris, S. (2000). Sky-high platforms propel telecoms to

new heights. Wireless Europe, pages 33–37.

Holland, D. (1967). The Cardington rainfall experiment.

The Meteorological Magazine, 96(1140):193–202.

Mannes, T., Fischer, C., Randeu, W., Kock, K., and Schon-

huber, M. (2002). Optimised satellite communica-

tions using weather radars. In 20th AIAA International

Communication Satellite Systems Conference and Ex-

hibit, Montreal, Quebec, Canada.

Mariño, P., Sigüenza, C., Poza, F., Vázquez, F., and

Machado, F. (2003a). Data warehouse ﬁlters and

power plant failure prediction models. In 7th IASTED

International Multi-Conference, Power and Energy

Systems, Palm Springs, California, USA.

Mariño, P., Sigüenza, C., Poza, F., Vázquez, F., Machado,

F., and Lamela, H. (2003b). Predictive maintenance

measurement system. In International Signal Process-

ing Conference (ISPC) & Global Signal Processing

Expo (GSPx), Dallas, Texas, USA.

Moore, R., Jones, D., Cox, D., and Isham, V. (2000). Design

of the HYREX raingauge network. Hydrology and

Earth System Sciences, 4(4):523–530.

Niemczynowicz, J. (1988). The rainfall movement - a valu-

able complement to short-term rainfall data. Journal

of Hydrology, 104:311–326.

Redaño, A. and Lorente, J. (1993). Modelling the spatial

and temporal distribution of rainfall intensity at local

scale. Theoretical and Applied Climatology, 47:25–

32.

Vilar, E., Montanari, C., Pontes, M., Mello, L. S., Lorente,

J., and Codina, B. (1997). Scattering and extinction:

Dependence upon raindrop size distribution in tem-

perate (Barcelona) and tropical (Belem) regions. In

10th International Conference on Antennas and Prop-

agation, number 436, pages 2230–2233, Heriot-Watt

University, Edinburgh, UK.

ICETE 2004 - WIRELESS COMMUNICATION SYSTEMS AND NETWORKS