Whats the fastest way to translate sqlite data to Google Chart format?

I have a sqlite database with the following scheme

CREATE TABLE `sensor` (
    `id`    TEXT NOT NULL UNIQUE,
    `description`   TEXT
);
CREATE TABLE `temperature` (
    `time`  INTEGER NOT NULL,
    `sensor`    TEXT NOT NULL,
    `value` REAL,
    FOREIGN KEY(`sensor`) REFERENCES sensor(id)
);

with temperature values from different sensors in it. One Entry looks like

unixtimestamp|sensor-id|temperature
1459802024|sensor1|22.0

I want to plot the data with the google Chart library (line plot). The javascript ajax-calls the following php-script:

<?php 

ini_set("display_errors", 1);
//header('Content-Type: application/json');
$db = new SQLite3('/file/to/sqlite/db');

$result = $db->query('SELECT * FROM sensor');
$sensors = array();    
$dataTable = array();



$interval = filter_input(INPUT_GET,'interval',FILTER_VALIDATE_INT, array("options"=>array("default" => time(),"min_range"=>0)));
$starttime = time() - $interval;

while($row = $result->fetchArray(SQLITE3_ASSOC)){
    $sensors[$row['id']] = $row['description'];
}

$et = microtime(true) - $_SERVER['REQUEST_TIME_FLOAT']; 
echo "</br>get available sensors: $et </br>";

$dataTable['cols'][] = array('id' => 'time_axis', 'label' => 'Time', 'type' => 'datetime');
$offset = 0;
foreach($sensors as $id => $description) {
    //Add Line for Sensor
    $new_column = array('id' => $id, 'label' => "$description ($id)", 'type' => 'number');
    $dataTable['cols'][] = $new_column;
    // Define your SQL statement //
    $stm = $db->prepare('SELECT time, value FROM temperature WHERE sensor=:sensor AND time >= :starttime ORDER BY time');
    $stm->bindValue(':sensor',$id,SQLITE3_TEXT);
    $stm->bindValue(':starttime',$starttime,SQLITE3_INTEGER);

    $result = $stm->execute();
    $et = microtime(true) - $_SERVER['REQUEST_TIME_FLOAT']; 
    echo "</br>sql sensor $offset: $et </br>";

    while($row = $result->fetchArray(SQLITE3_ASSOC)){  
        $date = $row['time']*1000;
        $temp = $row['value'];
            $cell = array();
            $cell[] = array('v' => "Date($date)");
            for ($i = 1; $i <= $offset; $i++) {   
                $cell[] = array('v' => null);
            }
            $cell[] = array('v' => $temp);           
            $dataTable['rows'][]['c'] = $cell;
        }
    $et = microtime(true) - $_SERVER['REQUEST_TIME_FLOAT']; 
    echo "</br>complete sensor $offset: $et </br>";
        $offset++;
}
$json =  json_encode($dataTable);
$et = microtime(true) - $_SERVER['REQUEST_TIME_FLOAT']; 
echo "</br>done: $et </br>";
?>

I observed that iterating through the rows takes up to 1 seconds. There are ~7k entries total and the code runs on a raspberry pi.

get available sensors: 0.0048329830169678 
sql sensor 0: 0.067873001098633 
complete sensor 0: 1.0311329364777 
sql sensor 1: 1.0555880069733 
complete sensor 1: 1.2514669895172 
done: 1.4120419025421 

Is there anything i could do to reduce computation time significantly ?

Reference for the data format: http://ift.tt/1Xd8NuB

http://ift.tt/RSFcv4