The State of Western Australia v Clarke - [2019] WASC 162

JURISDICTION : SUPREME COURT OF WESTERN AUSTRALIA
IN CRIMINAL

CITATION:   THE STATE OF WESTERN AUSTRALIA -v- CLARKE [2019] WASC 162
CORAM:   CORBOY J
HEARD:   14, 15, 19 & 22 MARCH 2019
DELIVERED          :   22 MARCH 2019
PUBLISHED           :   15 MAY 2019
FILE NO/S:   INS 82 of 2012
BETWEEN:   THE STATE OF WESTERN AUSTRALIA
Applicant
AND
AMBROSE JOHN CLARKE
First Accused
XAVIER GERARD CLARKE
Second Accused

Catchwords:

Nil

Legislation:

Criminal Procedure Act 2004 (WA), s 98(2)

Result:

Application to exclude evidence dismissed
Observations made on the limitations and use of telecommunications data made

Category: B

Representation:

Counsel:

Applicant	:	Mr P M Usher & Mr A C Ebell
First Accused	:	Mr S M Brennan, Ms A M Padmanabham & Ms A White
Second Accused	:	Mr J A Davies & Dr C J Hampson

Solicitors:

Applicant	:	Director of Public Prosecutions (WA)
First Accused	:	AP Legal
Second Accused	:	Legal Pathways

Case(s) referred to in decision(s):

The State of Western Australia v Coates [2007] WASC 307
Ward v R [2018] VSCA 80

CORBOY J:

The application

By indictment dated 21 June 2013, Ambrose Clarke and Xavier Clarke were jointly charged that on 30 May 2011 at Perth they murdered Peter John Davis. They were convicted of the charge on 3 December 2013 following a trial (the First Trial).
They appealed their conviction and on 8 February 2018, the Court of Appeal allowed the appeal, set aside the judgment of conviction and ordered a new trial. The new trial commenced on 26 March 2019.
The State's case against each of the accused is circumstantial. One set of circumstances relied on by the State concerns the location of Mr Davis and the accused on 30 May 2011. The State seeks to prove their movements on that day in several ways, including through evidence of the location and coverage of telecommunication cell sites (radio base stations) against which various mobile phone communications registered ('pinged').
By application dated 24 January 2018 (the Application), Xavier Clarke applied for orders pursuant to s 98(2) of the Criminal Procedure Act 2004 (WA) (CPA):
An order that the evidence of:
(a)Salvatore Stillone and;
(b)Jasmin Watson;
comprising:
(c)expert opinion as to the probable geographical coverage of telephone cell towers and;
(d)inferences that may be drawn from such expert opinion with respect to the location of various telephones of the accused and others at various times, to be relied upon by the Prosecution, listed in the schedule hereto ("objectionable inferences"),
(together, the "Objectionable Evidence");
be excluded as inadmissible, or alternatively, excluded as unfair to the accused in the exercise of the judicial discretion.
Ambrose Clarke joined in the Application. The Application was opposed by the State. Xavier Clarke subsequently advised that his objection only concerned the evidence proposed to be given by Mr Stillone.
Mr Stillone is the Senior Manager, State Radio Planning Quality, WA for SingTel Optus Pty Ltd (Optus). He gave evidence at the First Trial.[1] A number of exhibits were also tendered through him: P239 ‑ P252. The exhibits comprised maps depicting predicted telecommunications coverage for various Optus radio base stations (also referred to as 'cell sites' or 'telecommunications towers') located in the Perth Metropolitan Area.
[1] ts 2981 ‑ 3053 (18 November 2013).
The schedule of 'objectionable inferences' annexed to the Application referred to various paragraphs from a draft outline of the State's opening provided to the court and the parties. The relevant paragraphs anticipated that Mr Stillone would give evidence in the new trial about telecommunications data generated by the Optus mobile network that was similar to the evidence he gave in the First Trial.
The Application was supported by a report provided by Professor Coutts, Emeritus Professor of Telecommunications at the University of Adelaide. The State filed and served in the Application two further statements by Mr Stillone. Those statements provided further details about the Optus network and the data captured by the network. Mr Stillone also provided a letter to the DPP in which he responded to questions concerning aspects of Professor Coutts' report. Mr Stillone and Professor Coutts each gave evidence at the hearing of the Application.
Xavier Clarke contended that the evidence of Mr Stillone should be excluded on grounds that:
(a)Mr Stillone is not specially qualified to hold and express the opinions that form the Objectionable Evidence and lacks sufficient technical knowledge of the scientific basis of the subject matter of his opinions, including in relation to the 'large error rates in cell signal area propagation';
(b)the Objectionable Evidence incorrectly asserts a greater degree of reliability with respect to inferences about the geographical location of a mobile phone than that which is established by science;
(c)there are no scientific validation studies, by a field test or other means, of the laboratory software and measurement techniques used to predict radio base station coverage to establish that they are sufficiently reliable to ground inferences as to geographical location;
(d)the laboratories from which radio base station research is undertaken are not accredited in accordance with ISO/IEC 17025;[2]
(e)the inferences sought to be led about the location of a mobile phone within a geographical area of the size asserted tend to reverse the onus of proof by requiring the accused to provide an innocent explanation of an unproven fact.
[2] ISO/IEC 17025 'General Requirements for the Competence of Testing and Calibration Laboratories'. The standard has been adopted by Standards Australia and the National Association of Testing Authorities Australia.
I delivered oral reasons for dismissing the Application on 22 March 2019 and indicated that I would also publish settled written reasons. I concluded that Mr Stillone is qualified to express opinions concerning the radio signal coverage of various Optus radio base stations located in the Perth Metropolitan Area. I further concluded that he should be permitted to give evidence about that matter in the trial. However, the evidence presented in the Application, including the evidence given by Mr Stillone in the First Trial, established that there are limitations inherent in the way in which the coverage of a radio base station is predicted. Mr Stillone recognised and, in my view, is qualified to give evidence about those limitations. The State's case at trial must necessarily acknowledge the limitations.

The State's allegations

The State's case as foreshadowed in its draft outline of opening may be briefly summarised as follows. It is alleged that:
(1)Mr Davis was seriously assaulted by the accused sometime between 6.20 am and 7.50 am on 30 May 2011. It is likely that the assault occurred at or near 7 Townsend Street, Malaga, the address of a business called 'Full of Fun'.
(2)Mr Davis sustained multiple head and facial injuries consistent with multiple blunt force traumas that caused him to suffer a cardiac arrhythmia. The accused wrapped Mr Davis in black plastic and placed him in the cargo area of his four-wheel drive vehicle. Mr Davis was alive but unconscious when that occurred. Mr Davis' vehicle was then driven to the rear carpark of the Great Eastern Motor Lodge, Great Eastern Highway, Rivervale.
(3)Mr Davis and his wife, Jennifer Davis, were the effective owners of two businesses operated by Topseal Concrete Services Pty Ltd and TopAdditives Australia Pty Ltd ('Topseal' and 'TopAdditives' respectively). Topseal provided concrete sealing and related services.
(4)Ambrose Clarke and Mr Davis were in a business relationship between about 2006 and mid-2009. In late 2006/early 2007, Ambrose Clarke purchased a franchise from Topseal Concrete Services. He also invested a considerable sum in TopAdditives Australia.
(5)The business relationship between Ambrose Clarke and Mr Davis ended acrimoniously. Ambrose Clarke claimed he was owed approximately $300,000 by Mr Davis as a result of their business dealings. Mr Davis denied that he owed any money to Ambrose Clarke.
(6)Ambrose Clarke commenced operating another business under the name 'Purecoat'. In late 2009, he formed a partnership to operate the business with Mr Spiteri. Purecoat operated in competition with Topseal.
(7)Ambrose Clarke made various attempts to recover the money he claimed he was owed by Mr Davis. Those attempts included engaging Mr Ioasa to recover the money.
(8)Mr Ioasa was associated with a group of men who were collectively referred to as 'the Estonians'. Mr Ioasa, Ambrose Clarke and the Estonians were involved in incidents that were intended to intimidate Mr Davis and members of his family.
(9)Sometime in early May 2011, Mr Spiteri concluded that Ambrose Clarke had defrauded the Purecoat business. On 13 May 2011, Mr Spiteri terminated Ambrose Clarke's access to the bank accounts, telephones, computers, vehicle fuel cards and other operating items used by the business. He also advised Ambrose Clarke that his involvement with the business was at an end.
(10)Ambrose Clarke was in dire financial trouble by May 2011. He blamed Mr Davis for his financial difficulties.
(11)Following the termination of his relationship with Ambrose Clarke, Mr Spiteri arranged with Mr Davis for Topseal to provide Purecoat with plant and equipment and to assist in undertaking work. Ambrose Clarke was further aggrieved by Mr Davis' involvement in completing work for Purecoat.
(12)As at mid-late May 2011, Ambrose Clarke used a mobile phone with service number 0431 082 892 (the 0431 phone) and drove a blue Toyota Hilux vehicle (the blue Hilux). He was familiar with the suburb of Malaga as he had worked at Bunnings, Malaga between 1995 and mid‑2006.
(13)Xavier Clarke worked as a truck driver. His work place was located on Horrie Miller Drive, Perth Airport/Kewdale.
(14)On 22 May 2011, Ambrose Clarke purchased a Virgin mobile phone with a service number 0468 528 819 from Affinity Allphone, Kingsway (the 0468 phone). On the same day, he contacted a Virgin mobile customer service number and enquired about setting up and adding credit to a pre-paid SIM card. Later that day, he activated the 0468 phone using a SIM card starter pack and giving his name as 'Albert Norrish'. He also gave a false address and a contact telephone number that was connected to an apartment owned by a friend, Adrian Lee. Later still, he purchased a pre-paid mobile phone voucher.
(15)On 23 May 2011, Ambrose Clarke again called a Virgin mobile customer service number and made further enquiries using the name Albert Norrish. He then used the 0431 phone to call 'Full of Fun'. No legitimate purpose has been identified for that call.
(16)Mr Davis had two mobile phones ‑ a phone that he ordinarily used for business and personal calls (the business phone) and a phone that was primarily used while he was on holiday (the Bali phone). On 24 May 2011, three calls were made to Mr Davis' business phone by someone using the 0468 phone. The calls were not answered.
(17) On 26 May 2011, Xavier Clarke searched Google Maps for Townsend Street, Malaga. He then zoomed in on 7 Townsend Street. On the same day, someone called and spoke to Mr Davis using the 0468 phone. The State alleges that the purpose of the call was to lure Mr Davis to the vicinity of 7 Townsend Street, Malaga.
(18)On 29 May 2011, Ambrose Clarke spoke by phone with an acquaintance, Mr Reid. On the evening of 29 May 2011, Ambrose Clarke went to Mr Reid's home and cut a large piece of black plastic from a larger amount of plastic stored at Mr Reid's house. Ambrose Clarke told Mr Reid that he required the black plastic for a 'small job'. He also asked Mr Reid if he could provide him with some protective clothing but Mr Reid declined the request.
(19)Between 5.44 am and 6.01 am on 30 May 2011, the 0431 phone registered off a sector (antenna) at the Optus Maylands Peninsula radio base station. At 6.01 am, Ambrose Clarke called Xavier Clarke using the 0431 phone. Xavier Clarke's mobile phone registered off a Rivervale radio base station.
(20)At about 6.10 am on 30 May 2011, Mr Davis left his family home in High Wycombe in his four‑wheel drive vehicle. He drove to a delicatessen in High Wycombe. CCTV footage shows him dressed in work clothes. His vehicle was subsequently recorded by CCTV cameras travelling in Malaga, including on a road heading in the direction of Townsend Street. The GPS system in Mr Davis' vehicle was programmed for travel to 7 Townsend Street, Malaga.
(21)At 7.12 am, Mr Davis' son, Kurt Davis, sent a text message to his father's business phone. There was no reply to the text. The business phone registered off a sector at the Lockridge radio base station.
(22)At about 7.33 am, Xavier and Ambrose Clarke spoke by phone. Xavier Clarke's phone registered off the Rivervale North radio base station and Ambrose Clarke's phone registered off a sector at the Maylands Peninsula radio base station. The State alleges that Mr Davis was met with 'foul play' between 6.01 am and 7.33 am.
(23)At about 7.50 am, an employee of the Great Eastern Motor Lodge noticed Mr Davis' vehicle parked in the rear carpark of the Lodge.
(24)Shortly afterwards, Mr Davis' wife, Jennifer Davis, called her husband's business phone. The call went through to voicemail. The business phone was registering off a sector at the Belmont radio base station at the time of the call. The phone continued to register off that radio base station until it was moved by police on 31 May 2011.
(25)Between 9.21 am and 11.21 am, Mr Davis' Bali phone was registering off a sector at the Malaga radio base station. The phone and other property belonging to Mr Davis were still in the Malaga area at this time, suggesting that a third person may have been involved in the murder of Mr Davis.
(26)At 9.33 am, a text message was sent from the Bali phone, registering off the Malaga radio base station, to Jennifer Davis stating that Mr Davis was 'takin a day off'. The text message, the use of the Bali phone and the possibility of taking a day off work were all inconsistent with Mr Davis' normal practices. Also, the text message must have been sent by someone who knew the identity of Mr Davis' wife.
(27)During the morning of 30 May 2011, the 0431 phone registered off sectors located at radio base stations at Mount Lawley South, Highgate, Claisebrook, Bibra Lake, Murdoch and Leeming. In the afternoon, the 0431 phone registered off sectors located at radio base stations in Highgate, Mount Lawley, Como, Bull Creek, Murdoch, Coogee, Spearwood, Kardinya and Leeming.
(28)Members of Mr Davis' family, friends and work colleagues commenced searching for Mr Davis during the morning of 30 May 2011. They called Mr Davis' business phone and Bali phone many times during the day and that night. At 1.40 pm, Jennifer Davis reported Mr Davis as a missing person to the police.
(29)At 10.09 pm on 30 May 2011, the 0431 phone registered off the Kewdale South radio base station. Ambrose Clarke called Xavier Clarke, who was still working at the time. Xavier Clarke finished work and left his place of work on Horrie Miller Drive at 11.59 pm.
(30)Mr Davis' son, Kurt Davis, and his then girlfriend, Natasha Klanjscek, were looking for Mr Davis on the night of 30/31 May 2011. They drove along Great Eastern Highway, entering the carpark of a motel and a travel lodge in search of Mr Davis' vehicle. They saw a blue Hilux utility parked adjacent to a silver coloured hatchback in Toorak Road, Rivervale. Kurt Davis thought that the vehicle was similar to the blue Hilux driven by Ambrose Clarke. Two people were seated in the vehicle; Kurt Davis identified Ambrose Clarke as sitting in the front passenger seat. He did not recognise the person sitting in the driver's seat. A third person was at the rear of the silver hatchback. That person was dressed in high visibility work clothes and the State's case is that the blue Hilux belonged to Ambrose Clarke and the silver hatchback belonged to Xavier Clarke.
(31)Kurt Davis decided to investigate why Ambrose Clarke was parked in Toorak Road. He drove onto Great Eastern Highway with the intention of driving around the block and back onto Toorak Road. He inadvertently drove into the carpark of the Great Eastern Motor Lodge. He spotted Mr Davis' vehicle parked in the carpark. He called family members to advise he had located his father's vehicle.
(32)While Kurt Davis was speaking to his family, the blue Hilux entered the carpark of the Great Eastern Motor Lodge. The vehicle stopped directly in front of the vehicle driven by Kurt Davis. The blue Hilux then began to reverse. Kurt Davis moved his vehicle forward to block the blue Hilux from leaving the carpark. The blue Hilux rammed Kurt Davis' vehicle repeatedly and was eventually able to manoeuvre past Kurt Davis' vehicle and onto Great Eastern Highway.
(33)At about 12.21 am on 31 May 2011, the blue Hilux was recorded by CCTV cameras on a street in Rivervale driving at speed towards Great Eastern Highway. The vehicle was subsequently intercepted by police officers for a traffic breach. The police noted damage to the rear of the vehicle. By this time, Ambrose Clarke was the driver and sole occupant of the vehicle. He told the police officers he had been involved in an accident and was on his way to report the accident.
(34)At 12.22 am, Kurt Davis made a 000 call. Police arrived at the Great Eastern Motor Lodge approximately 20 minutes later. Other members of Mr Davis' family had arrived at the Lodge by this time. The police entered Mr Davis' vehicle and discovered his body wrapped in black plastic.
The balance of the police investigation and the circumstances relied on by the State to allege that Ambrose and Xavier Clarke murdered Mr Davis are not relevant to the Application other than to note that on 31 May 2011 the police conducted a search of Ambrose Clarke's house in which they located, among other things, the 0431 phone and the 0468 phone. The police also located items relating to the purchase and activation of the 0468 phone.

Mr Stillone's qualifications

Mr Stillone has a degree in Electrical Engineering from the University of New South Wales, having graduated in 1989. He specialised in telecommunications. He also has a graduate diploma in Management from the University of Western Australia. He commenced working in the telecommunications industry in 1990.
Mr Stillone has been employed by Optus since 2000. He has been the Manager, State Radio Planning Quality since January 2005. He is responsible for the planning and design of the Optus global system for mobile (2G), the universal mobile telecommunications system (3G) and the long‑term evolution (4G) networks in Western Australia. He has been involved in network planning and radio base station design throughout the time that he has worked in the telecommunications industry.
Mr Stillone has attended training provided by the software vendor of a network coverage prediction tool or model utilised by Optus. The relevant tool is the Teoco Asset 'Microcell3' tool (the Teoco Asset Tool). The tool is based on the Okumura‑Hata 'model' of radio propagation and has been in use throughout the time Mr Stillone has been employed by Optus.
The Okumura‑Hata model is an empirically derived equation that describes a logarithmic relationship between radio signal strength and distance. Optus uses a prediction model or tool to predict signal strength and network coverage for radio base station sectors (the Prediction Model). The model has four components ‑ radio propagation, building clutter, terrain and site data. Building clutter, terrain and site data are obtained from separate databases. It was not clear from the evidence whether the Teoco Asset Tool comprised software that incorporated the four components for the Prediction Model (that is, whether the Teoco Asset Tool was the Prediction Model) or whether the Teoco Asset Tool only provided the radio propagation data for the model.

I will continue to refer to the entire model used by Optus to predict network sector coverage (including for the purpose of generating probable and possible coverage maps for individual radio base station sectors) as the Prediction Model. I will refer to the radio propagation component of the Prediction Model as the RP Model. The RP Model is based on the Okumura‑Hata model of radio propagation. I inferred from the evidence that there are separate prediction models for the 2G, 3G and 4G networks.

The evidence given by Mr Stillone in the First Trial

In summary, Mr Stillone stated at the First Trial that:
(a)Optus operates 2G, 3G and 4G mobile phone networks with radio base stations for each network.
(b)Typically, a radio base station contains three antennas, which are referred to as sectors. The combined effect of the three sectors is to provide 360 degrees coverage at a site.
(c)A mobile phone that is switched on but not active (that is, not being used to make or receive a call) will 'camp' on the sector that has the strongest radio signal compared to other sectors. That sector will not necessarily be closest to the mobile phone as various factors can interact to determine the strength of a signal ‑ for example, building and vegetation clutter, terrain and the technical capacity of radio base stations.
(d)The records maintained by Optus for billing purposes include the sector on which a communication registered (the sector the call 'pinged' off). The records also identify whether a call was made on the 2G or 3G network.
(e)The information encoded in a mobile phone enables a network to constantly know the location of a phone. This permits the registration of a phone to shift seamlessly between sectors as the phone moves. However, the call records maintained by Optus only identify the sector on which a call first registered ‑ that is, the records do not identify other serving sectors if the sector on which the communication registered changes during a call. Further, the call records do not indicate where a mobile phone was located when a call is terminated.
(f)Optus maps the area within which each sector in a radio base station is predicted to be the most dominant ‑ that is, the sector with the strongest signal relative to other sectors. The maps are developed using a predictive model (that is, the Prediction Model) that takes into account radio propagation, building clutter, terrain and site data. The area within which a sector is predicted to be the most dominant relative to other sectors is referred to as the 'probable' coverage area. An area of 'possible' coverage can also be mapped. That is an area within which it is predicted a radio base station sector will have sufficient signal strength to sustain a phone call.
(g)The boundaries of a coverage area are not finite; there may be small movements of the boundaries.
(h)The 2G network has a feature that redirects a signal to the next strongest sector if the dominant sector is too busy. The 3G network does not possess that feature ‑ a call will fail where a sector has no capacity to receive a further communication. However, the 3G network is more 'sophisticated' when a phone is active as it permits simultaneous connection to multiple cells.[3]
(i)Exhibits P244 ‑ P252 comprised network maps showing predicted areas of probable and possible coverage for various radio base station sectors.
(j)Calls may be registered on a sector from a mobile phone that is located outside the predicted probable coverage area for that sector. Mr Stillone estimated that 'maybe five percent tolerance, you know, of the cell boundary, in terms of the traffic maybe that can be served outside that boundary compared to the total traffic captured by the whole cell'.[4]
(k)Occasionally, interference from outside sources can affect a radio base station. The weather can also affect the physical propagation of a radio signal and the coverage area in a 3G radio base station can reduce to compensate for high volumes of traffic. Other radio base stations will then take up the excess traffic. This phenomenon is known as 'cell breathing'.[5]
(l)Cell breathing is more likely to occur with a mobile phone that is located at the edge of a coverage area and might affect about five percent of total traffic for a busy sector. The orientation of the mobile phone can affect which surrounding sector registers when cell breathing occurs. The positioning of the phone in relation to a sector can also affect coverage ‑ in particular, a mobile phone located towards the back of a sector might register off another sector.
(m)The Prediction Model incorporates certain assumptions ‑ that the relevant sector and all surrounding sectors are working properly and that the sector is not 'overloaded' or 'congested'. Further, the propagation of radio waves across water will be different to propagation across a land mass. That difference is independent of the effect of building and vegetation clutter.
[3] ts 3023 (18 November 2013).
[4] ts 3026 (18 November 2013).
[5] ts 3037 (18 November 2013).
Mr Stillone also gave detailed evidence in the First Trial about the probable and possible coverage areas for particular radio base station sectors. The prosecution invited the jury to draw certain inferences about the location of Ambrose Clarke, Xavier Clarke, Mr Davis and Mr Davis' mobile phones during 30/31 May 2011 from the coverage maps and Optus call charge records. The prosecution also relied on that material to challenge parts of Ambrose Clarke's evidence concerning his location at various times on 30 May 2011.
There was no challenge in the First Trial to the admissibility of Mr Stillone's evidence. Further, his evidence was not disputed by counsel for Ambrose and Xavier Clarke (as distinct from contesting the inferences to be drawn from the evidence). Rather, the cross‑examination of Mr Stillone was directed at clarifying aspects of his evidence and emphasising matters that were relevant to the inferences, if any, about location that could be drawn from the network coverage maps and call records maintained by Optus.
Mr Stillone was not asked in the First Trial to give a detailed explanation of the Prediction Model or its components, including the RP Model. The primary focus of his evidence was on explaining how a mobile device communicates with a radio base station sector and the interpretation of the probable and possible coverage maps that were tendered. His evidence also dealt with factors that could affect the reliability of any inference to be drawn from the maps.

Mr Stillone's further statements

The two further statements made by Mr Stillone for the purpose of the Application provided additional detail about:
(a)records maintained by Optus of communications between its network and a subscriber whose device (mobile phone, handset, tablet or dongle) was in idle mode;
(b)records maintained by Optus of the distance between a subscriber's device and a radio base station;
(c)various Optus radio base stations.
In summary, Mr Stillone stated:
(a)The provision of mobile telecommunication services requires Optus to maintain a network of radio base stations that use antennas and radio waves to communicate with mobile devices. The radio base stations transmit and receive radio signals between Optus' network and a subscriber. The sectors of a radio base station are configured to provide coverage within a particular area.
(b)A sector consists of an omni‑directional antenna. Ordinarily, a radio base station consists of three sectors; each sector is separately numbered 1 to 3.
(c)Communications between the Optus network and a mobile device that is in idle mode (that is, the device is switched on but inactive) are stored in a database known as the Visitor Location Register (VLR).
(d)A mobile device regularly communicates with its subscriber network. This enables a record to be maintained of the serving base station that applied to a device at a specified time. A device will transmit data to the Optus network about the strength of signals received from a serving base station and neighbouring base stations. As a device is moved, the network may switch a device from one serving base station to a different serving base station depending on reported changes in signal strength.
(e)The Optus network maintains measurement records that give the distance between a device and a base station.
Mr Stillone annexed to his statements maps identifying the location of various radio base stations and details for those stations: station name, cell ID, property address, coordinate system (latitude and longitude), technology (2G/3G/4G) and antenna direction in degrees by reference to true north.

Mr Stillone's letter dated 1 March 2019

Mr Stillone provided further information concerning the Prediction Model in a letter to the DPP dated 1 March 2019. He confirmed that the model incorporates clutter and terrain databases. As with the RP Model, those databases were obtained by Optus from third party suppliers. The clutter and terrain databases include obstruction height details and clutter loss offsets. Clutter heights are added to terrain heights to perform obstruction loss calculations as part of the prediction process used by the Prediction Model. The terrain database includes terrain height contours (that is, heights above sea level) of the landscape across an area. Radio diffraction is modelled and accounted for in predicting coverage and the Prediction Model also includes a radio base station database which includes data on antenna heights, antenna types, base station types and power.
Mr Stillone further stated in his letter that:
(a)Consistent with industry practice, the Prediction Model has been calibrated using drive test data. A 'tuning' exercise was undertaken in the late 1990s for the 2G model and in 2005 for the 3G model. The tuning exercise involved collecting coverage data from a carrier wave test transmitter set up across different radio environments. The signal strength of the test transmitter was measured using a vehicle that drove a set test route. The coverage data was then incorporated into the Prediction Model and the model 'tuned' until error was minimised to a satisfactory value.
(b)Optus has continued to use network drive test data to fine tune models in various radio environment conditions.
(c)'Anomalous propagation' is a radio concept also known as 'fading'. It occurs as a result of fluctuations in the received signal power mainly due to variations in the signal power amplitude and signal phase because of the different objects in a radio signal's path. The objects attenuate a received signal's power as they absorb or scatter part of the signal. Radio base stations and handsets use coded techniques or algorithms to mitigate the effects of fading. Optus makes allowance for fading in the Prediction Model by adding a safety margin to the service level coverage prediction plots.
(d)Coverage predictions are not 'one hundred percent' accurate. The Prediction Model generates a prediction based on certain formulae and data. It is well understood in the telecommunications industry that no mobile coverage tool will be completely accurate. However, in Mr Stillone's experience, the Prediction Model provides a good indication of the service areas of a radio base station for the purpose of identifying the area within which a call can be reliably made on the Optus network. The Prediction Model is, in Mr Stillone's opinion, fit for – and is used for – service planning and optimisation and for making investment decisions about the network.

Professor Coutts' report

Professor Coutts was provided with the following information for the purpose of preparing his report:
(a)the coverage maps tendered in evidence at the First Trial ‑ exhibits P240 to P251 and P252‑1 to P252‑42;
(b)the transcript of evidence given by Ambrose Clarke and Mr Stillone at the First Trial;
(c)the prosecutor's opening and closing addresses;
(d)the trial judge's directions to the jury.
Professor Coutts was further instructed that Ambrose Clarke claims he received prank calls during the period relevant to the charge. The calls were shown as 'missed calls' and the telephone numbers of the callers have been matched to numbers recorded in the telephone directory. Professor Coutts indicated in his report that it was impossible to ascertain from call records whether missed calls were prank calls.
Professor Coutts did not contend that inferences could not be validly drawn about the location of a person using telecommunications data. Rather, his report focussed on the expertise required to interpret such data and the limitations in its use ‑ the certainty with which inferences could be made about the precise location of a mobile phone by reference to call charge records and coverage maps. He observed that '[t]he 'science' regarding what can be inferred from the cell ID in a CCR record is still under study and any inference based on seemingly complex software methods should be treated with caution and done by an expert who has knowledge of radio propagation'.[6]
[6] Professor Coutts' report, 3.4.
Professor Coutts considered that a prospective witness should satisfy four requirements to qualify as an expert on telecommunication data. The witness should possess:
(a)knowledge, acquired by experience, about the information available from mobile networks;
(b)knowledge of, and experience in, the complexities of radio propagation particularly around 1‑2GHz;
(c)knowledge of radio coverage in a mobile environment;
(d)an understanding of the strengths and weaknesses of propagation prediction tools.
Professor Coutts added:
[s]ome, but not all of this knowledge can be gained from tertiary training. The expertise is mainly acquired while working as a radio field‑test engineer or radio optimiser dealing mainly with how cell coverage and traffic load are handled by various network technologies.[7]
[7] Professor Coutts' report, 3.2.
It is not necessary to consider the extent to which Professor Coutts' criteria for expertise in telecommunications data should be accepted. Mr Stillone's experience is as a radio engineer dealing with signal coverage, traffic loads and radio base station design and operation for the Optus network. Professor Coutts accepted that Mr Stillone satisfied the first three requirements to qualify as an expert witness on telecommunications data.[8] However, he considered that Mr Stillone did not possess sufficient knowledge of the 'weaknesses of RF software propagation to accurately predict coverage and the need for a measurement validation process'.[9]
[8] Mr Stillone was cross-examined to establish that he had not engaged in academic research into radio propagation. An expert may, of course, be qualified to express opinions about a subject through experience rather than study. In my view, it was not necessary for Mr Stillone to have undertaken studies on radio propagation to qualify as an expert having regard to the subject matter of his evidence. That view accorded with the emphasis that Professor Coutts placed on experience in his discussion of the qualifications for expertise.
[9] Professor Coutts' report, 3.2.
It was apparent from the balance of Professor Coutts' report that this reservation reflected a concern with the use that had been made of Mr Stillone's evidence in the First Trial rather than with a lack of expertise. Early in his report, Professor Coutts observed that, '[t]his case in my experience over 10 years as an expert in Australia and currently in Canada is the most over reach in interpretation that I have seen'.[10] Professor Coutts continued '[t]his report will discuss the underlying reasons why such software predictions and coverage maps are prone to error'.[11] Consistent with that approach, Professor Coutts stated that the object of his report was to 'support the more broad opinion that inference of location from cell ID obtained from Call Charge Records … should be interpreted by an appropriate expert and not regarded as a form of scientific evidence due to the complexity of radio propagation and the limitations of software coverage prediction tools'.[12] It was apparent from Professor Coutts' report, and his evidence in the Application, that the reference to 'scientific' evidence was intended as a warning that telecommunications data cannot be used, by itself, to definitively determine the location of a mobile phone or its user. As will be explained, that is not the purpose for which telecommunications data is produced by a mobile service provider such as Optus and Mr Stillone did not suggest otherwise in his evidence in the First Trial or in the Application.
[10] Professor Coutts' report, 3.4.
[11] Professor Coutts' report, 3.4.
[12] Professor Coutts' report, 2.
In summary, Professor Coutts stated in his report that:
(a)It is important that the influence of local buildings (clutter) is taken into account when predictions over small distances are required to be accurate. Accurate radio frequency predictions require detailed clutter and height data and a prediction model needs to account for the impact of possible reflections from buildings and water in or near a city such as Perth.
(b)Field measurements are required from time to time to confirm the accuracy of predicted signal paths and to tune a prediction model.
(c)Software coverage prediction models are intended for radio coverage planning and not for predicting a specific mobile location, especially where a single call is relied on as evidence of location.
(d)Anomalous propagation due to short term meteorological changes can affect the accuracy of predicted signal coverage. Line of sight components, building obstructions and water proximity can also affect signal predictions.
(e)The actual propagation characteristics of a radio base station antenna may be required to examine the possibility of a call being received close to the back of the recorded sector.
Professor Coutts identified in his report two examples of what he regarded as misuse of telecommunications data in the First Trial. The first example concerned whether Ambrose Clarke had travelled along Guildford Road early on the morning of 30 May 2011. He was alleged to have made calls on the 0431 phone between 5.44 am and 6.01 am that morning. The State further alleged that he was in Rivervale when the calls were made, while Ambrose Clarke maintained that he was on Guildford Road. The calls registered off a sector located at the Optus Maylands Peninsula radio base station. The probable coverage area for the relevant sector did not extend to Guildford Road. However, the possible coverage area for the sector did include parts of Guildford Road. As Professor Coutts observed, the limitations on the use of coverage maps to identify the location of a mobile phone meant that the possibility that Ambrose Clarke was travelling on Guildford Road at the time that the calls were made could not be excluded solely by the telecommunications data presented by the State.
The second example concerned a call made from the 0431 phone at 8.33 am on 30 May 2011. Mr Clarke gave evidence that he made the call while travelling to Newcastle Street from a location in the CBD. The call registered off a sector located on a building in Melville Parade, South Perth (close to the Narrows Bridge end of the Freeway). Again, Professor Coutts noted that the possible coverage area for the relevant sector extended across the river and into the CBD and beyond. Accordingly, the possibility that Ambrose Clarke was travelling on or near Newcastle Street at the relevant time could not be excluded by the telecommunications data.

Professor Coutts referred in his report to papers he had co‑authored with Mr Hugh Selby on the use of mobile phone data as evidence of location: 'The Safe and Unsafe Use of Mobile Phone Evidence';[13] 'Problems with cellphone evidence tendered to "prove" the location of a person at a point in time';[14] and 'Mobile Ping Data – Metadata for Tracking'.[15] The first two papers contained what the authors described as basic information about mobile networks and communications and the data generated from the operation of a network. The focus of the papers was on probable and possible coverage maps and the use of coverage prediction software for geolocation purposes. Each paper contained a discussion of the limitations of using coverage maps as a means of identifying the location of a person at any point in time. The third paper discussed the use of VLR and home location register data as location evidence.
[13] Paper delivered at Communications Policy Research Forum, September 2008 and Public Defenders Criminal Law Conference, 2009.
[14] (2016) 13 Digital Evidence and Electronic Signature Law Review 76.
[15] (2017) 14 Digital Evidence and Electronic Signature Law Review 22.
An attachment to the paper entitled 'The Safe and Unsafe Use of Mobile Phone Evidence' summarised information that had been provided to Professor Coutts and Mr Selby about the use of mobile phone location evidence. Reference was made in the attachment to the judgment of Blaxell J in The State of Western Australia v Coates which contained observations on the limitations of telecommunication data as evidence of location.[16] His Honour's comments were described by Professor Coutts and Mr Selby as the 'high water mark' of judicial discussion and were commended to lawyers participating in hearings in which mobile phone evidence was to be adduced.
[16] The State of Western Australia v Coates [2007] WASC 307.
It is to be noted that Blaxell J did not suggest in Coates that telecommunications evidence was inadmissible on any of the grounds advanced by Xavier Clarke. Rather, his Honour identified various factors which he considered were relevant to assessing the evidence.

The oral evidence in the Application

It is not necessary to set out in detail the oral evidence given by Mr Stillone and Professor Coutts at the hearing of the Application. It is sufficient to note the following matters.
First, Mr Stillone identified various features of a mobile network and the Prediction Model that, in his opinion, were relevant to the use of network coverage maps and call charge records as evidence of location. His evidence on those matters was consistent with the evidence he gave in the First Trial and in his statements and letter to the DPP. That evidence has been outlined above. Nevertheless, it may assist the parties in this trial if the features identified by Mr Stillone are summarised in a list:
(a)The Prediction Model and the coverage predictions generated by the model are not designed for the purpose of identifying the location of a mobile phone at any particular time. The purpose of the Prediction Model is to assist in optimising the operation of Optus' mobile network ‑ to monitor and enhance the performance of the network and to enable technical and investment decisions to be made about the network and its numerous components.
(b)The boundaries of the probable and possible coverage areas for a sector are a 'moving feast'[17] – that is, small changes to the boundaries can occur due to factors such as the volume of traffic ('cell breathing').
(c)The network is designed so that a mobile phone, whether active or idle, will be transferred seamlessly between sectors as the phone travels. Consequently, there will be some overlap between boundaries to prevent calls dropping out.
(d)There is some tolerance or 'error rate' inherent in the terrain and clutter databases. The databases are produced from satellite images or aerial photographs.
(e)It is possible for a signal to be picked up from the rear of a sector. However, Mr Stillone considered that this rarely occurred.
(f)A phone can be serviced by a particular sector even though it was located outside the probable coverage area for that sector and the extent to which a phone will register on the closest sector can vary from sector to sector.
(g)There are various assumptions inherent in the RP Model – for example, that the nearest sector is not overloaded with traffic and that all radio base stations are operating correctly.
(h)The building clutter database was updated by Optus in about March 2013. The database utilised by Optus prior to 2013 was created in the early 1990s. Changes to the clutter and terrain databases may have been made during the model tuning exercise that occurred in 2005. However, building clutter will change over time – for example, building density may increase in areas through high-rise construction.
(i)The line of sight to a sector may affect the distance a signal can travel. Accordingly, a signal will travel further over water.
(j)The call charge records maintained by Optus do not indicate whether particular sectors were either congested or inoperative at the time a call was made.
(k)Some devices such as TV boosters can interfere with a signal if the device is operated near a radio base station. Interference will impede coverage and will tend to affect a number of radio base stations within a localised area. Moreover, depending on the extent of the interference and the area affected, interference may impact only on the voice quality of a call or it may prevent a call from being made.
[17] ts 4442 (14 March 2019).
Second, the factors identified by Mr Stillone are in addition to the matters Professor Coutts referred to in his report as summarised above. Professor Coutts disclaimed expertise in software prediction models for network coverage. However, he did claim knowledge of the factors relevant to assessing the accuracy of data generated by such models for the purpose of determining location. That understanding had been acquired through research as an academic and industry consultant and I accept that Professor Coutts does possess expert knowledge about that matter.
Third, Mr Stillone gave further details of his training in the use of the Prediction Model. He stated that his training had exposed him to all functions and features of the model, including how to produce coverage maps. His training also gave him an overview of the RP Model.
Fourth, Mr Stillone described the Okumura‑Hata model as logarithmic. Professor Coutts described the model as empirical. Although counsel did not further explore those descriptions with Mr Stillone and Professor Coutts, some insight into the model was provided by two references mentioned in Professor Coutts report.[18] It is not proposed to further describe the model given the limited evidence provided in the Application other than to note that the model predicts signal strength across distance in an urban environment according to a logarithmic regression equation which has as its variables frequency, radio base station effective antenna height above ground, mobile station antenna height above ground and distance. I infer that the co-efficient values for the model's variables were derived from field tests (so that the model is both logarithmic, as Mr Stillone stated, and empirical, as Professor Coutts stated). According to Professor Coutts, those field measurements were undertaken in Tokyo in the late 1970s.
[18] Hata, M 'Empirical Formula for Propagation Loss in Land Mobile Radio Services', IEEE Transactions on Vehicular Technology, VT‑29(3) (August 1980) and International Telecommunication Union, Recommendation ITU‑R p.1546‑5, 'Method for point-to‑point area predictions for terrestrial services in the frequency range 30 MHz to 3000 MHz'.
Fifth, Mr Stillone stated that radio frequency engineers employed by Optus gained considerable local knowledge through working in the field undertaking, for example, drive testing. Local knowledge plays a significant part in making decisions about the network – its design and optimisation. In addition, adjustments are made from time to time to radio base stations. The site database incorporated in the Prediction Model is amended as those changes are made.
Sixth, Mr Stillone explained that validation of the Prediction Model occurs for different radio signal environments. There are several models that can be tuned to reflect differences in signal environments. In addition, from time to time Optus verifies that the predictions generated by the Prediction Model are within acceptable limits of accuracy for different environments and locations using test drive data collected by scanners.
Seventh, Mr Stillone explained that Optus no longer provide coverage maps to law enforcement officers. That is because Optus can now provide more precise information about the distance between a mobile device and a radio base sector against which the device has registered. However, that information is not available for a period as far back as 2011.
Eighth, it was accepted by counsel for Xavier Clarke that Mr Stillone had given frank evidence about the Prediction Model and the limitations to using the model's predictions for the purpose of identifying the specific location of a mobile phone. In my view, that concession was rightly made having regard to his evidence and the evidence of Professor Coutts.

Disposition

It was clear that Mr Stillone had a good understanding of the components of the Prediction Model and its application for the purpose of predicting the coverage area for a sector located at a radio base station operated by Optus. The reservation that Professor Coutts expressed about Mr Stillone's expertise reflected a concern over the use that had been made of his evidence in the First Trial and the forensic decisions of counsel about the issues to be explored in his evidence, rather than a lack of knowledge about the limitations of using telecommunications data for the purpose of determining the location of a mobile phone. There was no substantive difference in the opinions expressed by Mr Stillone and Professor Coutts on the derivation and application of network coverage models such as the Prediction Model or on the limitations of using such models to generate data from which an inference about location can be drawn.[19] Those are matters which are properly the subject of expert evidence and about which, in my view, Mr Stillone is qualified to give evidence.
[19] Mr Stillone's evidence was entirely consistent with the discussion of telecommunications data in the papers written by Professor Coutts with Mr Selby. Mr Stillone and Professor Coutts differed in their estimates of how far coverage might extend beyond the predicted boundaries of a coverage area. However, in my view that was a difference in expert opinions that did not affect the admissibility of the evidence that the State proposed to adduce through Mr Stillone.
In his closing submissions, counsel for Xavier Clarke contended that the coverage maps produced by Optus 'do not have a sufficient degree of tested and proven reliability to ground the inferences that the State seeks to propound as to the specific task of the location of particular phones at particular times as identified by the call charge records'.[20] The contention rested on a number of further propositions:
•The coverage maps were not designed for that purpose but are intended to be used for optimising the operation of the Optus mobile network.
•It was not possible to 'safely' conclude that a call originated in the probable coverage area for a sector. The use of coverage maps for the purpose of fixing the location of a phone has not been researched and tested and the degree of uncertainty inherent in predicted coverage has not been scientifically quantified.
•Unlike other forms of forensic evidence such as DNA analysis, the maps were not produced pursuant to a rigorous testing regime by an entity that complied with standards for conducting scientific or forensic testing.
•Factors relevant to determining the location of a mobile phone at a particular time are not known or recorded – for example, whether sectors were inoperative or affected by transient interference.
•The use of the word 'probable' to describe one set of maps connotes a degree of certainty that may be misleading.
•The maps are an abstraction as demonstrated by the linear boundaries that define the coverage areas.
•The pictorial nature of the maps, coupled with expert evidence about their derivation and interpretation, may result in the jury placing undue weight on the maps.
[20] ts 4536 (19 March 2019).
The fact that the coverage maps have not been produced for a forensic purpose, or by an entity that has agreed to apply standards that are intended for laboratories and other facilities that undertake testing and calibration activities, does not mean that they are inadmissible. Evidence of a technical nature is not confined to evidence that has been obtained pursuant to a recommended standard for facilities that specialise in testing. The common law rules of evidence are ultimately concerned with reliability. The question is whether the evidence to be adduced is sufficiently reliable to be admissible.
The coverage maps are produced by the Prediction Model. The Prediction Model incorporates the RP Model, which is based on the Okumura‑Hata model of radio propagation. As has been explained, the Okumura‑Hata model is an empirical model that was developed in the late 1970s and, on the evidence presented in the Application, reflects the physics of radio waves. The model has been widely used in the telecommunications industry. The wide-spread use of the model indicates that it must have been found to be a reliable predictor of the relationship between radio signal strength and distance.
The Prediction Model also incorporates data that accounts for factors that affect radio propagation. It was not suggested that the incorporation of that data was controversial. Rather, the evidence indicated that the derivation and application of the Prediction Model reflected industry practice in Australia and elsewhere.
The Prediction Model has been used by Optus since at least 2000 and probably from before that time (it has been in use for the entire time that Mr Stillone has been employed by Optus and a prediction tool was used by Telstra when he was employed by that entity prior to 2000). The Prediction Model has been tuned in the past and is subject to validation by field testing. Allowance is made in network coverage maps for phenomena such as anomalous propagation. Although the Predication Model has not been developed for forensic purposes, it has been used by Optus for many years to make technical and business decisions about the operation of its mobile network. It may be readily inferred that the Prediction Model has been found to provide reasonably accurate coverage predictions, given its purpose and the length of time over which it has been utilised. The limitations of the Prediction Model and the coverage maps are reasonably well defined and understood.
In my view, the data generated by the Prediction Model in the form of the coverage maps is sufficiently reliable to be admissible. That, of course, is subject to the intended use of the maps; their use must properly reflect their limitations. In that regard, I accept that there were instances in the First Trial where the inferences sought to be drawn from the coverage maps did not fully reflect the workings of the Optus mobile network and the limitations inherent in the coverage maps. As Professor Coutts indicated in his report, this most obviously occurred where there was a failure to acknowledge the possibility that a call may have been made from outside the probable area of a sector's predicted coverage. That did not occur throughout the trial, but there were instances where the limitations in the evidence were not fully acknowledged. There was also a tendency at times to treat the coverage maps as being more than merely predictions.
The limitations of the telecommunications data to be adduced by the State must be recognised in the use that is made of the evidence. That does not mean the data lacks probative value. The data generated by the Optus network, at least for the period relevant to this case, cannot fix the location of a phone at particular point in either the probable or possible coverage areas for a sector. However, subject to certain matters relating to area boundaries, it appears on the evidence presented in the Application that it is highly likely a phone will be within the probable or possible coverage area of the sector it registers against ‑ that is, the phone must be sufficiently close to the sector the phone is registered to, at least, sustain a signal. That will be so even if the call has been switched from another sector for some reason.
On the evidence presented in the Application, the actual boundaries of a coverage area at a particular time may vary from the boundaries depicted in a coverage map due to the 'tolerance' or 'error' permitted for a boundary and phenomena such as cell breathing and interference. However, the effect of Mr Stillone's evidence was that those factors were likely to produce only small variations in the actual boundaries of a coverage area from those depicted in a coverage map. Professor Coutts suggested the variation may be greater than indicated by Mr Stillone, but even on his evidence the variations were not so great as to render telecommunications data inutile.
Accordingly, the telecommunications data proposed to be adduced by the State might provide evidence from which an inference can be drawn about the general location of a phone at a time when it registers off a sector ‑ that is, the general area within which the phone was located. If multiple calls have been made from a phone, the data might also permit an inference to be drawn about the direction the phone has travelled (although that will be subject to the possibility of 'switching' between sectors). The significance of any inference will depend on the purpose for which it is sought to be drawn. For example, it might be relied on merely to infer that a person was within a broadly defined area at a particular time. Alternatively, it might be relied on as one circumstance, to be considered with other evidence, from which an inference can be drawn more precisely about the location of a person.
In short, the use that is sought to be made of the telecommunications data will be critical. I do not consider that admitting evidence of the data will undermine the right of the accused to a fair trial, provided that the evidence is not misused. However, the use of the evidence will remain subject to the overriding duty to ensure a fair trial.
I also do not consider that the nature of the evidence is such that a jury is likely to be misled, provided that the evidence is presented in a way that reflects its limitations. As the decision of the Victorian Court of Appeal in Ward v R[21] indicates, this extends beyond the use made of the evidence to the way in which the opinions of a telecommunications expert, such as Mr Stillone, are expressed. For example, the expert evidence should not be put in a way that conveys the erroneous impression that:
[21] Ward v R [2018] VSCA 80.
(a)the coverage maps are more than predictions;
(b)inferences about a phone's location can be drawn from telecommunications data without due regard to the limitations inherent in the data; or
(c)the probability of a call being made inside or outside the predicted areas of coverage can be, or has been, quantified (unless there is a proper basis for asserting that the probability can be reliably quantified).

I certify that the preceding paragraph(s) comprise the reasons for decision of the Supreme Court of Western Australia.
MB
Associate to the Honourable Justice Corboy
15 MAY 2019