Unpacking data structure with Python

This should get you started. Unpack the received bytes using struct.unpack, then do some parsing for fields that are represented by less than 1 byte of data.

from struct import unpack

# data recieved, 88 random bytes for example purpose
data = b'\xdb[\x91wdI\t\xef\xc6c\xde\x14\xac\x1e\x08\x10.f\xc0\xbd\xfd\xa15\x8cP\x101\xed\xc5\xd9\x98X\xb5\xc2\x00Z\xd2\xb9\xb0Xa\x04\xfa\xb8\xceA\x94_7\xc7\xde\t\xf2kX\x9d2\xc3\x84\xb3\x19\x8e\xf5\x99\xc3\xba\x08\xaa0$\x17\xfbd\xbb\x7f\xfd&\xf5\x1aU\t`\x11@zD\xce\xff'

# unpack the struct into variables
(
    abcde, fw_ver, cur_layer, fs_radix,  # 0 needs parse
    fghij, knpl, fbg_thermistor,  # 1 needs parse
    tx_ambient_temp, reserved2,  # 2
    num_ffpi_peaks, num_fbg_peaks,  # 3
    num_dut2_peaks, num_dut1_peaks,  # 4
    num_dut4_peaks, num_dut3_peaks,  # 5
    reserved7, qr, acq_counter3,  # 6 needs parse
    serial_number,  # 7
    kernel_timestamp_microseconds,  # 8
    kernel_timestamp_seconds,  # 9
    kernel_src_buffer, kernel_buffers,  # 10
    error_and_kernel_rt_loc0,  # 11 needs parse
    header_length, header_ver, buffers,  # 12
    dut2_gain, dut1_gain,  # 13
    dut4_gain, dut3_gain,  # 14
    dut2_noise_thresh, dut1_noise_thresh,  # 15
    dut4_noise_thresh, dut3_noise_thresh,  # 16
    hw_clk_div, peak_data_rate_div,  # 17
    granularity,  # 18
    reserved4,  # 19
    starting_lambda,  # 20
    ending_lambda  # 21
) = unpack(
    '>'  # big endian
    'BBBB'  # 0 needs parse
    'BBH'  # 1 needs parse
    'HH'  # 2
    'HH'  # 3
    'HH'  # 4
    'HH'  # 5
    'BBH'  # 6 needs parse
    'I'  # 7
    'I'  # 8
    'I'  # 9
    'HH'  # 10
    'I'  # 11 needs parse
    'HBB'  # 12
    'HH'  # 13
    'HH'  # 14
    'HH'  # 15
    'HH'  # 16
    'HH'  # 17
    'I'  # 18
    'I'  # 19
    'I'  # 20
    'I',  # 21
    data
)

# 0 parse abcde
acq_triggered = bool(abcde & 0x80)
calibration_fault = bool(abcde & 0x40)
start_of_frame = bool(abcde & 0x20)
primary_fan_state = bool(abcde & 0x10)
secondary_fan_state = bool(abcde & 0x08)
s0_mux_state = bool(abcde & 0x04)
s1_mux_state = bool(abcde & 0x02)
s2_mux_state = bool(abcde & 0x01)

# 1 parse fghij
xfer_type = fghij >> 4
soa_therm_limit = bool(fghij & 0x08)
soa_current_limit = bool(fghij & 0x04)
tec_over_temp = bool(fghij & 0x02)
tec_under_temp = bool(fghij & 0x01)

# 1 parse knpl
operating_mode = knpl >> 6
triggering_mode = (knpl & 0x30) >> 4
sm041_mux_level = (knpl & 0x0c) >> 2
sw_position = knpl & 0x03

# 6 parse qr
nrz_command = qr >> 5
reserved6 = qr & 0x1f

# 11 parse
error = error_and_kernel_rt_loc0 >> 24
kernel_rt_loc0 = error_and_kernel_rt_loc0 & 0xffffff

I assumed big endian since this is coming over TCP, but that might be wrong. If all the data looks off, try using < for little endian. Or if you're unlucky some values might be one while some are the other, in which case you'll have to split this into multiple unpacks. You'll also have to further process these values, as I'm sure some of them aren't supposed to be interpreted as ints.

The unpack format can be shortened to

(all, those, variables) = unpack('>6B9H2BH3I2HIH2B10H4I', data)

but I don't think it's as clear.