以個人經驗.. iOT 用在地震服務以 A20-OLinuXino-LIME2 收集相關資料，當然用 Arduino 傳相關資料丟 WebAPI 當然也可以..

1. 用 C# 收集 A20 設備的 CPU、RAM、HD
2. 用 JSON 自訂集合相關格式
3. 透過 C# 接 http WebAPI 來將其資訊寫到資料庫

Arduino 的部份就在裡面用 URL 只要傳的出去應該都可以。

如果你是用 SQL Server 2016 版以上，WebAPI 接到資料後將 JSON 直接丟給 SQL SERVER 後用 SQL JSON QUERY 來拆解再存起來，非常簡單..

針對「如果收到的一段16進制的資料，該怎麼切割欄位」，
提供 Python 可能做法以供參考。
(主要是自己太閒又有興趣練習～)
(## ■ 基本範例 前的不用認真～ 只是無聊的初始化～)

## Python==3.8.16;
## 
import random
## get random bytes; (@ cpython/Lib/random.py)
def randbytes(n): return random.getrandbits(n * 8).to_bytes(n, 'little')
## 
## (myHL01b, myHL01b, myHL01b): highlight for print();
myHL01b = '\033[36m'; myHL01e = '\033[0m'
myHL02b = '\033[33m'; 
## myMsg for print();
myMsg01 = lambda na, obj : '' \
  + str(na) \
  + ': type:[' + myHL01b \
  + str(type(obj)) \
  + myHL01e + '], Value:[' + myHL01b \
  + str(obj) \
  + myHL01e + '];'
myMsg02 = lambda na, bs : myMsg01(na, bs.hex('-').upper())
## 
## ■ 基本範例
print(myHL02b + '　■ Basic Show:' + myHL01e)
myBytesRa = randbytes(20)
myTN='myBytesRa'; print(myMsg02(myTN, locals()[myTN]))
## 從字串轉取 Bytes 物件。
## https://docs.python.org/3/library/stdtypes.html#bytes.fromhex
myBytes01str = '03 4C DF 48 4C 48 03 DF 03 4C DF 48'
myBytes01 = bytes.fromhex(myBytes01str)
myTN='myBytes01'; print(myMsg02(myTN, locals()[myTN]))
print(myMsg02('myBytes01 slicing', myBytes01[2:4]))
## Bytes 轉取 int
## https://docs.python.org/3/library/stdtypes.html#int.from_bytes
myInt01 = int.from_bytes(myBytes01[2:4], byteorder='big', signed=True)
myTN='myInt01'; print(myMsg01(myTN, locals()[myTN]))
## 
## □ struct 範例
## https://docs.python.org/3/library/struct.html
print(myHL02b + '　□ "struct" Show:' + myHL01e)
import struct
## read: big-endian ( short integer (size:2),)
print(myMsg02('myBytes01 slicing', myBytes01[2:4]))
myInt02 = struct.unpack('>h',myBytes01[2:4])
myTN='myInt02'; print(myMsg01(myTN, locals()[myTN]))
## read: big-endian (short integer (size:2), float (size:4))
print(myMsg02('myBytes01 slicing', myBytes01[2:8]))
myInt03 = struct.unpack('>hf',myBytes01[2:8])
myTN='myInt03'; print(myMsg01(myTN, locals()[myTN]))
print(myMsg01('myInt03[0]', myInt03[0]))
print(myMsg01('myInt03[1]', myInt03[1]))
## 
print('')
## 
## ■ numpy 範例
print(myHL02b + '　■ "numpy" Show:' + myHL01e)
import numpy as np
## Usage: myMsg03 for print();
myMsg03 = lambda st, dt : 'dtype(' + st + '):[' + myHL01b + str(dt) + myHL01e + '];'
## dtype: https://numpy.org/doc/stable/reference/arrays.dtypes.html
## frombuffer: https://numpy.org/doc/stable/reference/generated/numpy.frombuffer.html
print(myMsg03(
  "np.dtype(np.int16).newbyteorder('>')", 
  np.dtype(np.int16).newbyteorder('>')
))
myDT01 = np.dtype('>i2') 
print(myMsg02('myBytes01 slicing', myBytes01[2:6]))
myNPAry01 = np.frombuffer(myBytes01[2:6], dtype=myDT01)
myTN='myNPAry01'; print(myMsg01(myTN, locals()[myTN]))
print('')
## Structured-dtype(?):
## Show for dtype-WebDoc: [(field_name, field_dtype, field_shape), ...]
print(myMsg03(
  "np.dtype(np.float32).newbyteorder('>')",
  np.dtype(np.float32).newbyteorder('>')
))
print(myMsg02('myBytes01 slicing', myBytes01[:8]))
myNPAry02 = np.frombuffer(myBytes01[:8], dtype=[
  ('f1', '>i2'), ('f2', '>i2'), ('f3', '>f4')
])
myTN='myNPAry02'; print(myMsg01(myTN, locals()[myTN]))
print('')
## 
## Show for dtype-WebDoc: {'names': ..., 'formats': ..., 'offsets': ..., 'titles': ..., 'itemsize': ...}
## Sliced myBytesRa[:16] for [ValueError: buffer size must be a multiple of element size]
print(myMsg02('myBytesRa slicing', myBytes01[:16]))
myNPAry03 = np.frombuffer(myBytesRa[:16], dtype={
  'names' : ['f1','f2','f3'], 
  'formats' : ['>i2', '>i2', '>f4']
})
myTN='myNPAry03'; print(myMsg01(myTN, locals()[myTN]))
print(myMsg01('myNPAry03[1]', myNPAry03[1]))
print(myMsg01('myNPAry03[1][0]', myNPAry03[1][0]))
## 

至於存資料庫，找 drivers 連線、寫入，我覺得應該有一堆範例了：

• PSQL Wiki 驅動程式列表
• PSQL Wiki Python 相關
  (Wiki 內有部分過時資訊，需自行判斷～)

這裡 還有人討論最佳方法呢！(不過是針對 numpy 就是了～)